|Home | About | Journals | Submit | Contact Us | Français|
The year 2017 marks the twentieth anniversary of terpenoid cyclase structural biology: a trio of terpenoid cyclase structures reported together in 1997 were the first to set the foundation for understanding the enzymes largely responsible for the exquisite chemodiversity of more than 80000 terpenoid natural products. Terpenoid cyclases catalyze the most complex chemical reactions in biology, in that more than half of the substrate carbon atoms undergo changes in bonding and hybridization during a single enzyme-catalyzed cyclization reaction. The past two decades have witnessed structural, functional, and computational studies illuminating the modes of substrate activation that initiate the cyclization cascade, the management and manipulation of high-energy carbocation intermediates that propagate the cyclization cascade, and the chemical strategies that terminate the cyclization cascade. The role of the terpenoid cyclase as a template for catalysis is paramount to its function, and protein engineering can be used to reprogram the cyclization cascade to generate alternative and commercially important products. Here, I review key advances in terpenoid cyclase structural and chemical biology, focusing mainly on terpenoid cyclases and related prenyltransferases for which X-ray crystal structures have informed and advanced our understanding of enzyme structure and function.
Terpenes, also known as terpenoids or isoprenoids, comprise the most chemically and structurally diverse family of natural products. Currently numbering more than 80000 members in a greater family that also includes steroids and carotenoids, the terpenome accounts for nearly one-third of all compounds currently characterized in the Dictionary of Natural Products (http://dnp.chemnetbase.com).1,2 The structural complexity of this vast chemical library belies relatively simple biosynthetic roots: head-to-tail coupling reactions of 5-carbon precursors yield linear, achiral C5n isoprenoid diphosphates (n = 1, 2, 3, etc.) (Figure Figure11),3,4 which in turn undergo cyclization reactions to yield a myriad of products typically containing multiple fused rings and stereocenters (some examples are shown in Figure Figure22).5−19 These reactions are catalyzed by enzymes known as terpenoid synthases; terpenoid synthases that catalyze cyclization reactions are also known as terpenoid cyclases. Terpenoid cyclization reactions are the most complex reactions found in nature, in that on average more than half of the substrate carbon atoms undergo changes in bonding, hybridization, and stereochemistry during the course of a multistep cyclization cascade. Moreover, since cyclic terpenoids typically cannot be generated from linear precursors in the absence of an enzyme, the catalytic rate enhancement of a terpenoid cyclase over the uncatalyzed rate is immeasurably large.
General scheme of terpene nomenclature, linear precursors (OPP = diphosphate), synthase classification, and commonly observed domain architectures.
Chemodiversity is a hallmark of the branches of the terpenome family tree. Individual 5-carbon isoprenoid groups are color-coded to highlight their biosynthetic fates. Enzyme abbreviations: BPPase, bornyl diphosphate synthase; CPPase, chrysanthemyl diphosphate ...
Additional structural diversity in terpenoid natural products arises from the coupling chemistry employed to link C5 isoprenoid precursors. For example, C5 dimethylallyl diphosphate (DMAPP) and C5 isopentenyl diphosphate (IPP) can be linked in regular (i.e., head-to-tail) fashion to yield C10 geranyl diphosphate, which in turn can undergo condensation with additional IPP molecules to yield C15 farnesyl diphosphate (FPP), C20 geranylgeranyl diphosphate (GGPP), C25 geranylfarnesyl diphosphate (GFPP), and so on (terpenoid nomenclature and commonly used acronyms are summarized in Figure Figure11). Alternatively, two isoprenoid allylic groups can undergo irregular coupling reactions,20 such as cyclobutanation, branching, or cyclopropanation reactions, to yield alternative carbon skeletons that further diversify the terpenome (Figure Figure22).21,22
A hallmark of most reactions catalyzed by a terpenoid synthase is the cascade of multiple carbocation intermediates that define the reaction coordinate of catalysis. Such highly reactive intermediates could potentially alkylate and inactivate the enzyme, so the terpenoid synthase typically contains a nonpolar active site pocket that enables effective management and manipulation of these intermediates. However, this does not preclude the presence of occasional polar groups or solvent molecules in the terpenoid synthase active site (e.g., as observed in bornyl diphosphate synthase from Salvia officinalis and aristolochene synthase from Aspergillus terreus).23−25 Although such potential nucleophiles could rapidly quench a carbocation intermediate, they will not do so as long as they are fixed in positions where they are not properly oriented to react. Since some terpenoid cyclases generate hydroxylated products, reflecting quenching of the final carbocation intermediate by a water molecule, it is logical that these cyclases have evolved with suitable water management strategies to prevent premature quenching of carbocation intermediates.
Quantum mechanical studies of terpenoid cyclase reaction mechanisms indicate that tertiary carbocation intermediates are much more common than secondary carbocation intermediates.26−28 Additionally, these studies show that many carbocation intermediates in terpenoid cyclization cascades are subject to electron delocalization (e.g., through hyperconjugation and other through-bond coupling) that influences and directs the cyclization trajectory. Carbocation intermediates can be stabilized in the active site of a terpenoid cyclase by weakly polar interactions involving charge–charge, charge–dipole, and charge–quadrupole interactions with suitably oriented amino acid side chains. In particular, carbocation intermediates can be stabilized by the substantial partial negative charge generated by the ring π electrons of the aromatic side chains of phenylalanine, tyrosine, and tryptophan through cation−π interactions (Figures Figures33 and and44).29−31 Such carbocation stabilization can generally be achieved without risk of alkylation and accordingly comprises a strategy for electrostatic stabilization of multiple transition states in the terpenoid cyclization cascade.
(A) An aromatic ring is an electronic quadrupole, with no net charge and no net dipole. (B and C) Electrostatic surface potentials for the side chain aromatic groups of phenylalanine, tyrosine, and tryptophan reveal that the faces of these aromatic residues ...
Optimized geometries and stabilization energies calculated for carbocation−π interactions with benzene and indole indicate that the π systems of the side chains of phenylalanine, tyrosine, and tryptophan are capable of stabilizing ...
Although there are interesting exceptions,32 including antibodies that catalyze isoprenoid-like cyclization reactions,33−35 terpenoid cyclases generally fall into two main classes depending on the chemical strategy for initial carbocation formation:8 a class I terpenoid cyclase utilizes a trinuclear metal cluster to trigger the ionization of an isoprenoid diphosphate substrate to yield an allylic cation and inorganic pyrophosphate, whereas a class II terpenoid cyclase relies on a general acid (an aspartic acid side chain) to protonate the terminal carbon–carbon double bond of an isoprenoid substrate to yield a tertiary carbocation. The first X-ray crystal structures of terpenoid cyclases revealed evolutionarily distinct α-helical folds for class I and class II enzymes.36−38 As classified by Oldfield and colleagues,39,40 the active site of a class I cyclase such as bacterial pentalenene synthase36 or plant epi-aristolochene synthase37 is located in the middle of an α-helical bundle designated the “α fold”, and the active site of a class II cyclase such as squalene-hopene cyclase38 is located at the interface of two α-helical domains designated “β” and “γ” (Figure Figure55). Structural homology and 23% amino acid sequence identity between the β and γ domains of squalene-hopene cyclase8 suggest a primordial gene duplication and fusion event with evolution of catalytic activity at the domain–domain interface, so the γ domain of a class II terpenoid cyclase can also be considered as a β′ domain. Similarly, gene duplication and fusion of an ancient 4-helix bundle protein yielded the α fold of a class I terpenoid cyclase (see section 2.1).41
Domain diversity: terpenoid synthase structures generally consist of α, β, and γ domains (blue, green, and yellow, respectively) in various combinations. The γ domain is an insertion domain between the first and second ...
The crystal structure of taxadiene synthase, a diterpene cyclase from the Pacific yew, was the first to reveal that a class I terpenoid cyclase could exist as a fusion of αβγ domains (the α domain is catalytically active and the βγ domain is an inactive vestige) (Figure Figure55),42 in accord with predictions based on bioinformatics and experimental studies.39,43 Since epi-aristolochene synthase adopts αβ domain architecture (Figure Figure55),37 domain architecture for a class I terpenoid synthase is thus found as α, αβ, or αβγ. Interestingly, the crystal structure of the class II terpenoid cyclase ent-copalyl diphosphate synthase reveals αβγ domain architecture with a functional active site at the βγ domain interface and an inactive, vestigial α domain.44 Thus, domain architecture for a class II terpenoid cyclase can be βγ or αβγ. The evolution of catalytic activity in the class I or class II active site is accordingly a function of the biosynthetic needs of the cell and the particular organic reaction mechanism required for catalysis. The crystal structure of abietadiene synthase from the grand fir was the first to reveal the structure of a bifunctional αβγ terpenoid synthase with catalytically active class I and class II active sites that catalyze tandem diterpene cyclization reactions,45 and the crystal structure of α-bisabolene synthase from the grand fir revealed that a class I sesquiterpene cyclase could also adopt αβγ domain architecture.46
The domain architectures summarized in Figure Figure55 thus represent the context for the vast majority of terpenoid cyclization reactions in biology. Each terpenoid cyclase is found with tertiary structure consisting of a core domain architecture of α, αβ, βγ, or αβγ, as summarized in Figure Figure11. Notably, αα domain architecture is also observed in certain bifunctional terpenoid synthases (see section 7). In terms of quaternary structure, dimers, tetramers, and hexamers are observed, examples of which are discussed throughout this Review.
The terpenoid cyclization reactions discussed in this Review focus on those that utilize linear isoprenoid diphosphate substrates with all-(E) (i.e., all trans) isoprenoid double bonds, as illustrated in Figure Figure11. Indeed, terpenoid cyclization products of trans-isoprenoid substrates account for the great majority of terpenoid natural products. However, it should be noted that some terpenoid synthases utilize or generate linear isoprenoid diphosphates containing (Z) (i.e., cis) double bonds.47−51 Perhaps the most prominent example is undecaprenyl diphosphate synthase, which generates a C55 isoprenoid that functions in bacterial cell wall biosynthesis.52 These cis-prenyltransferases adopt a completely different fold53 in comparison with trans-prenyltransferases and therefore evolved independently of the enzymes discussed in this Review. It is interesting, however, that the trans-prenyltransferase FPP synthase also generates minor amounts of cis isomers (i.e., products GPP or FPP containing a C2–C3 cis-double bond),54 and that a cis-FPP synthase has been identified in Mycobaterium tuberculosis.55 Moreover, the sesquiterpene cyclase epi-aristolochene synthase from tobacco can utilize (cis,trans)-FPP as a cyclization substrate to generate an alternative array of cyclization products,56 and certain class II diterpene cyclases can utilize nerylneryl diphosphate as a substrate.57 Such catalytic versatility enhances the chemodiversity of the terpenome.
The remainder of this Review includes much of the classic work underlying modern studies and applications of terpenoid biosynthesis yet also emphasizes studies that have emerged since my last Review on this topic.13 As prominent terpenoid synthase archetypes are discussed, common mechanistic themes emerge regarding the chemical strategy for initial carbocation formation, carbon–carbon bond-forming reactions in multistep cyclization cascades, the fidelity or promiscuity of the protein catalyst in chaperoning these cascades, and the chemical strategy for terminating the cyclization cascade by proton elimination or capture by solvent. To set the stage for understanding the structural and chemical biology of terpene cyclases, it is instructive to first review the coupling enzymes that generate the linear substrates for terpene cyclization reactions. These chain elongation reactions are catalyzed by prenyltransferases.
The era of terpenoid synthase structural biology began with the X-ray crystal structure of avian FPP synthase, which catalyzes the 1′-4 (i.e., head-to-tail) coupling of C5 isoprenoid units (as represented by the “trunk” of the tree of terpenoid diversity shown in Figure Figure22).58 Also referred to as a regular coupling reaction and more generally as a prenyltransferase reaction, the 1′-4 chain elongation reaction is the most important of all nine possible isoprenoid couplings found in nature (Figure Figure66) because it provides the linear achiral substrates for cyclization by terpenoid cyclases. All other couplings illustrated in Figure Figure66 are referred to as irregular coupling reactions. As summarized in this section, both regular and irregular coupling reactions are catalyzed by metal-dependent enzymes that exhibit the α fold of a class I terpenoid synthase. As such, the FPP synthase structure58 foreshadowed the first terpenoid cyclase structures that would follow three years later.36−38
Naturally occurring isoprenoid coupling patterns. Individual C5 isoprenoid units are black and blue, and bonds between them are shown in red. Terpenoid synthases catalyze coupling reactions that yield these connections. The 1′-4 (head-to-tail) ...
Farnesyl diphosphate synthase is the archetype of a regular coupling enzyme. This class I terpenoid synthase catalyzes the condensation of DMAPP and two molecules of IPP to yield FPP. The enzyme adopts a unique α-helical fold characterized by bundles of α-helices,58 later designated the class I terpenoid synthase fold,8,36 and more recently designated the α fold.39,40,59 FPP synthase is a homodimer in which the active sites are oriented in parallel fashion (Figure Figure77), with aspartate-rich metal ion binding motifs DDXXD located on the upper wall of the active site on helices D and H in each subunit. Earlier studies of the related coupling enzyme hexaprenyl diphosphate synthase and other prenyltransferases,60,61 as well as FPP synthase from different species,62,63 implicated these aspartate-rich motifs for a metal binding function in catalysis. This metal binding function was confirmed in the initial X-ray crystal structure determination, in which these motifs were observed to coordinate to samarium (Sm3+) ions used as heavy atom derivatives to phase the initial electron density map.58 Later, the crystal structure of F112A/F113S avian FPP synthase complexed with GPP (the product of the first coupling reaction with DMAPP and IPP) revealed the binding of 2 Mg2+ ions, each coordinated by the first and last aspartate in the first aspartate-rich motif D117DXXD (Figure Figure88).64 The second aspartate-rich motif does not coordinate to metal ions in this complex, which represents an incomplete enzyme–substrate complex for the second coupling reaction of GPP and IPP to generate FPP.
Crystal structure of avian FPP synthase, an isologous dimer of 44-kD subunits, was the first to reveal the α fold of a class I terpenoid synthase. The active site of each monomer opens toward the top and is partially capped by loops and additional ...
Stereoview of avian F112A/F113S FPP synthase complexed with GPP. The side chains of D117 and D121 in the first aspartate-rich motif coordinate to two Mg2+ ions (green spheres), which in turn are coordinated by the diphosphate group of the substrate (C ...
In FPP synthase from Escherichia coli, the first aspartate-rich segment has a two-residue insertion so that it appears as DDXXXXD; thus, there can be subtle variations in this sequence that nonetheless sustain metal-binding function. The X-ray crystal structure of this enzyme complexed with IPP and the unreactive analogue of DMAPP, dimethylallyl-S-thiolodiphosphate (DMSPP), shows that both aspartate-rich motifs and the thiolodiphosphate group of DMSPP coordinate to 3 Mg2+ ions; octahedral metal coordination polyhedra are completed by solvent molecules (Figure Figure99).65 It is now clear that 3 metal ions are required to provide the electrophilic driving force that triggers isoprenoid diphosphate ionization and initial carbocation formation in most if not all terpenoid synthases that adopt the α fold.66 Ligand binding causes conformational changes that serve to cap the active site, thereby protecting reactive carbocation intermediates from bulk solvent.
(a) Stereoview of the active site of FPP synthase from E. coli complexed with IPP (C = green, O = red, and P = yellow) and the unreactive substrate analogue DMSPP (C = brown and S = yellow) reveals the binding of a full complement of 3 Mg2+ ions (blue ...
These structures provide a framework for understanding earlier enzymological and mechanistic studies of FPP synthase indicating an ionization-condensation-elimination sequence for catalysis.3 Coordination to 3 metal ions and hydrogen bond interactions with 3 basic residues (arginine and/or lysine residues)66 triggers the ionization of the allylic diphosphate moiety of DMAPP to yield a tightly bound allylic carbocation-inorganic pyrophosphate ion pair (Figure Figure1010).67−69 Interestingly, the possibility of such an ion pair to initiate an ionization-condensation-elimination sequence, as opposed to a concerted condensation-elimination sequence, had first been proposed in mechanistic studies of squalene biosynthesis.70−72 Alkylation of the si face of the C3–C4 double bond of IPP with inversion of configuration at the C1 atom of DMAPP yields a tertiary carbocation intermediate,73 which then undergoes elimination of the pro-R proton at C2 to yield the chain elongation product GPP.74 The crystal structure of the E. coli FPP synthase-DMSPP-IPP complex reveals that the diphosphate group of DMSPP is suitably oriented to mediate the stereospecific proton elimination that terminates the coupling reaction.65 General base catalysis by coproduct inorganic pyrophosphate had previously been suggested for FPP synthase;3,75−78 multistep general base-general acid functions for inorganic pyrophosphate may also prevail for terpenoid cyclases, as initially considered79 based on structural analysis of aristolochene synthase from Penicillium roqueforti as well as other cyclases.80 In accord with the processive function of FPP synthase, GPP then shifts to allow the binding of another molecule of IPP and a second ionization-condensation-elimination sequence, which generate the final product FPP as shown in Figure Figure1010.
(A) FPP synthase is a processive enzyme that catalyzes the coupling of IPP molecules to a growing isoprenoid chain. Condensation of DMAPP and IPP yields GPP. In the second round of the chain elongation reaction, (a) ionization of GPP yields an allylic ...
Being a processive enzyme, FPP synthase requires a stop signal to ensure that the reaction sequence is terminated upon generation of the correct final product. The active site of FPP synthase provides a template for catalysis that serves as a “molecular ruler”. Mutagenesis and structural studies of avian FPP synthase64 reveal that the side chain of F113 buttressed by the side chain of F112 forms the base of the pocket that dictates product isoprenoid chain length. The F112A and F113S mutations deepen the pocket by 5.8 Å. When incubated with DMAPP and IPP, this double mutant generates an array of C20-C70 isoprenoid products. Thus, the newly formed base of the pocket in F112A/F113S FPP synthase can undergo slight structural changes so as to open up to bulk solvent, thereby enabling the formation of longer products. Conversely, the active site pocket can be made more shallow by introducing the A116W or N114W constrictions, which introduces significant selectivity for GPP synthesis.81 Random chemical mutagenesis of FPP synthase from Bacillus stearothermophilus similarly identify active site residues that, when mutated, lead to the generation of mixtures of GPP, GGPP, and FPP.82
Since the initial structure determination of avian FPP synthase nearly 25 years ago,58 crystal structures of coupling enzymes that generate GPP, GGPP, hexaprenyl diphosphate, and octaprenyl diphosphate have also appeared.83−95 These enzymes similarly adopt the α fold of a class I terpenoid synthase, and their active site pockets are shallower or deeper to direct the chain elongation reaction to form short-chain, medium-chain, or long-chain isoprenoid products. Analysis of these structures illuminates details regarding the molecular ruler governing product chain length, which provides a foundation for protein engineering experiments in which alternative isoprenoid products can be generated.87 In one particularly striking example, octaprenyl pyrophosphate synthase from Thermotoga maritima can be engineered to make a linear C95 isoprenoid, which is the longest product generated by any trans-prenyltransferase (Figure Figure1111).84
(a) A “molecular ruler” governs product chain length in short-, medium-, and long-chain prenyltransferases: bulky hydrophobic residues define the base of the pocket that serves as a template to direct product chain length. In general, ...
Notably, a recent structural analysis indicates that the core α fold of FPP synthase likely resulted from a gene duplication and fusion event involving a primordial 4-helix bundle protein (Figure Figure1212).41 In this analysis, which utilized the structure of FPP synthase from E. coli, the aspartate-rich metal-binding motifs on helices D and H are topologically equivalent. Given the utility of 4-helix bundles in the modular design of de novo functional proteins,96−102 their appearance in terpenoid synthases suggests that new avenues in terpenoid biosynthesis might be explored through the de novo design and assembly of 4-helix bundles.
Recently, Poulter, Jacobson, and colleagues have reported the development of a new computational bioinformatics-directed approach for the study of more than 5000 terpenoid synthases putatively assigned as trans-prenyltransferases.92 In this study, genomic and crystal structure data are used to predictively guide the discovery and functional evaluation of short-, medium-, or long-chain trans-prenyltransferases. Impressively, the prediction of function in new enzymes is 62% correctly predicted and 94% correctly predicted within plus or minus a single 5-carbon isoprene unit. Sequence similarity networks103,104 illustrate evolutionary relationships in Figure Figure1313.
(A) Sequence similarity network of prenyltransferases with BLAST e-value cutoff = 1e–50. (B) Sequence similarity network of prenyltransferases with BLAST e-value cutoff = 1e–70. Nodes are color-coded according to the product generated ...
Irregular coupling reactions, also known as nonhead-to-tail coupling reactions, yield the general coupling patterns shown in Figure Figure66. Strikingly different carbon skeletons result from alternative isoprenoid coupling reactions, and this bond-making variability contributes to the vast chemodiversity of the terpenome. The major irregular isoprenoid coupling reactions include c1′-1–2 and c1′-2–3 cyclopropanation reactions, c1′-2–3–2′ cyclobutanation reactions, and 1′-2 branching reactions. Both regular and irregular isoprenoid coupling reactions comprise the roots of terpenoid diversity as represented in Figure Figure22.
In an elegant set of experiments involving the design, preparation, and analysis of chimeric terpenoid synthases, Poulter and colleagues demonstrated that cyclobutanation, cyclopropanation, and branching reactions can be catalyzed by a metal-dependent class I terpenoid synthase.105 To construct these chimeras, the amino acid sequence of sagebrush FPP synthase was progressively substituted with the amino acid sequence of sagebrush chrysanthemyl diphosphate synthase (75%/96% amino acid sequence identity/similarity). Strikingly, while FPP synthase generates FPP almost exclusively and chrysanthemyl diphosphate synthase generates monoterpene cyclopropanation and branching products, FPP synthase-chrysanthemyl diphosphate synthase chimeras generate varying arrays of regular and irregular monoterpene coupling products, including cyclobutanation products, through dissociative electrophilic alkylation mechanisms (Figure Figure1414). Although the recent crystal structure of lavandulyl diphosphate synthase reveals a protein fold unrelated to the characteristic α fold of a class I terpenoid synthase, it is notable that this coupling chemistry can be catalyzed in two distinct protein scaffolds.106
Chimeric terpenoid synthases generated from FPP synthase (FPPase-M) and chrysanthemyl diphosphate synthase (CPPase-M) generate all four fundamental isoprenoid coupling products when incubated with (A) DMAPP and IPP or just (B) DMAPP. Products are color-coded ...
Irregular isoprenoid coupling reactions in biological systems are not limited to monoterpene biosynthesis. Consider the condensation of two FPP molecules to yield presqualene diphosphate (Figure Figure1515). This reaction is catalyzed by human squalene synthase in the cholesterol biosynthesis pathway; the second step of the reaction is reduction by NADPH to yield products squalene and inorganic pyrophosphate. The crystal structure of human squalene synthase reveals the core α fold of a class I terpenoid synthase.107 Despite relatively low amino acid sequence identity with other class I terpenoid synthases, conservation of the overall α fold and signature aspartate-rich metal binding motifs (appearing in slightly varied form as D80TLED and D219YLED) similarly indicates evolutionary divergence from a common ancestral class I terpenoid synthase (Figure Figure1616). Interestingly, Pandit and colleagues note that the kink in helix G of squalene synthase is conserved in other terpenoid synthases; this kink exposes the main chain carbonyl groups of V175 and A176 to the active site cavity, such that the negative electrostatic potential of the helix dipole may stabilize carbocation intermediates in catalysis.107 More recent structural studies of hedycaryol synthase complexed with nerolidol also suggest a role for the helix G kink in carbocation stabilization.108
Human squalene synthase (SQS) catalyzes the coupling of two FPP molecules to yield presqualene diphosphate, a critical intermediate in steroid biosynthesis. Dehydrosqualene synthase (CrtM) catalyzes the same reaction in the biosynthesis of the golden ...
Human squalene synthase, avian FPP synthase, bacterial pentalenene synthase, and plant 5-epi-aristolochene synthase share insignificant overall amino acid sequence identity, yet these enzymes share the class I terpenoid synthase fold with conservation ...
Related to the eukaryotic squalene synthase reaction is the prokaryotic presqualene diphosphate synthase reaction catalyzed by the coupling enzyme dehydrosqualene synthase in Staphylococcus aureus (Figure Figure1515).109 In this bacterial pathogen, the FPP coupling reaction is the first committed step in the biosynthesis of the golden pigment staphyloxanthin, which is responsible for the eponymous coloration of S. aureus (“aureus” is the Latin term for a gold coin used in the Roman Empire).110,111 The extended conjugated π system of staphyloxanthin confers protective antioxidant properties, capable of trapping reactive oxygen species that might otherwise kill the bacterium. Staphyloxanthin is thus a potent virulence factor, and inhibition of its biosynthesis would render S. aureus more susceptible to the oxidative countermeasures of the human immune response in the event of infection.
The crystal structure of bacterial dehydrosqualene synthase reveals a class I terpenoid synthase fold similar to that of human squalene synthase; the ternary complex of dehydrosqualene synthase with two molecules of farnesyl thiolodiphosphate (FSPP, an unreactive FPP analogue) reveals how the two coupling substrates bind in the active site: the diphosphate group of one FSPP molecule coordinates to a full complement of 3 Mg2+ ions that are also coordinated by three aspartate residues and one asparagine residue in the aspartate-rich metal-binding motifs (Figure Figure1717).109 This is the first view of precatalytic substrate binding for a cyclopropanation reaction in the active site of a terpenoid synthase.
(A) Dehydrosqualene synthase (CrtM) from S. aureus adopts the α fold of a class I terpenoid synthase. (B) Superposition of bacterial dehydrosqualene synthase (green) and human squalene synthase (yellow) reveals homologous structures. (C) Two molecules ...
Noting the structural and functional similarities between dehydrosqualene synthase and squalene synthase, Oldfield and colleagues suggest that inhibitors developed against squalene synthase in the development of cholesterol-lowering drugs might also inhibit dehydrosqualene synthase and render S. aureus defenseless against oxidative reagents, or more importantly, to the oxidative burst of the mammalian immune system.109 Incubation of S. aureus with various squalene synthase inhibitors does indeed inhibit generation of the golden staphyloxanthin pigment; treatment with the cholesterol-lowering agent BPH-652 (Figure Figure1818) makes the resulting unpigmented bacteria ~15-fold more susceptible to killing by 1.5% hydrogen peroxide. In an animal model of S. aureus infection, mice treated with BPH-652 by intraperitoneal injection contain 98% fewer surviving bacteria (Figure Figure1818). Thus, inhibition of the class I terpenoid synthase dehydrosqualene synthase comprises a new therapeutic strategy for the treatment of methicillin-resistant S. aureus infection.109,112−117
(A) Inhibition of dehydrosqualene synthase with 0–1000 mM BPH-652 blocks staphyloxanthin pigment biosynthesis in S. aureus (IC50 = 110 nM). (B) Administration of BPH-652 exhibits substantial bactericidal activity against S. aureus following intraperitoneal ...
Finally, it is interesting to note that the coupling of two FPP molecules to generate squalene requires a sequence of three separate steps catalyzed by three distinct enzymes in bacteria. Poulter and colleagues118 demonstrate that the first enzyme catalyzes the coupling of two FPP molecules to generate presqualene diphosphate, which is converted to hydroxysqualene by the second enzyme in the biosynthetic sequence. Bacterial squalene synthase then utilizes FADH2 to reduce hydroxysqualene to squalene. Thus, prokaryotic squalene biosynthesis contrasts with eukaryotic squalene biosynthesis in terms of the number of enzymes involved in the biosynthetic sequence as well as the cofactor utilized for redox chemistry [human squalene synthase107 requires NADPH (Figure Figure1515)].
Cytosolic aromatic prenyltransferases are enzymes that catalyze the transfer of isoprenoid chains to aromatic rings (e.g., phenyl, naphthyl, or indole rings) in biosynthetic reactions leading to hybrid natural products that serve critical functions in biological systems. For example, the biosynthetic origins of the antioxidant naphterpin, an isoprenoid-polyketide natural product, are rooted in 1,3,6,8-tetrahydroxynaphthalene and GPP (Figure Figure1919A). These substrates are coupled through an alkylation reaction reminiscent of a classic Friedel–Crafts reaction. The X-ray crystal structure of an aromatic prenyltransferase reveals a novel 10-stranded antiparallel β-barrel fold, termed a PT-barrel fold,119 with an overall topology consisting of five repeated αββα motifs (Figure Figure1919B). Accordingly, these enzymes are designated as ABBA prenyltransferases.120
(A) Prenylation of 1,3,6,8-tetrahydroxynaphthalene (THN) or flaviolin by GPP is a critical step early in the biosynthesis of the antioxidant naphterpin. Other hybrid isoprenoid-polyketides are generated in similar fashion using isoprenoid substrates DMAPP, ...
Many ABBA prenyltransferases require Mg2+ to trigger the ionization of the isoprenoid diphosphate group, forming inorganic pyrophosphate and the allylic cation that alkylates the aromatic ring of the cosubstrate. Some, such as dimethylallyl tryptophan synthase, do not require Mg2+ but instead utilize a broad array of hydrogen bond interactions to trigger substrate ionization.121 Regardless, ABBA prenyltransferases do not adopt the protein fold of a class I or class II terpenoid synthase, nor do they contain the signature aspartate-rich motifs that characterize a class I terpenoid synthase. The crystal structure of the bacterial prenyltransferase functioning in naphterpin biosynthesis complexed with the unreactive substrate analogue geranyl thiolodiphosphate (GSPP) reveals that a Mg2+ ion is coordinated by the thiolodiphosphate group of GSPP and only a single protein residue, D62.119
In contrast with cytosolic ABBA prenyltransferases, prenyltransferases belonging to the UbiA superfamily are polytopic membrane proteins that catalyze the first committed steps in the biosynthesis of ubiquinone, menaquinone (vitamin K), and other prenylated aromatic compounds. Archetypical members of this enzyme family include microbial UbiA and MenA, and their eukaryotic homologues COQ2 and UBIAD1, respectively; the entire superfamily has been recently reviewed (Figure Figure2020).122 UbiA was first discovered by Young and colleagues123 and catalyzes the coupling of polyprenyl diphosphates of varying lengths124 to the ortho position (i.e., C3) of 4-hydroxybenzoic acid. Members of the UbiA superfamily are characterized by two aspartate-rich metal-binding motifs, NDXXDXXXD and DXXXD,61,125−127 which are reminiscent of the signature metal-binding motifs of class I terpenoid cyclases.
Recent crystal structure determinations of archaeal UbiA homologues from Aeropyrum pernix (ApUbiA) and Achaeoglobus fulgidus (AfUbiA) reveal an overall α fold identical to that of the α domain of a class I terpenoid synthase.41,128 These structures reveal that the helical segments containing the aspartate-rich motifs form a cap domain that encloses the active site, thereby protecting reactive carbocation intermediates from premature quenching by bulk solvent (Figure Figure2121). Huang and colleagues41 show that this fold exhibits pseudo 2-fold symmetry, identical to that of farnesyl diphosphate synthase, suggesting gene duplication and fusion of a primordial 4-helix bundle protein with the evolution of catalytic function at the interface of the two bundle domains (Figure Figure2222).
Crystal structure of ApUbiA reveals the characteristic α fold of a soluble class I terpenoid synthase adapted for function as an integral membrane protein. The cap domain (pink) encloses the hydrophobic central cavity formed by transmembrane helices ...
Crystal structure of AfUbiA reveals the characteristic α fold of a class I terpenoid synthase, which exhibits pseudo-2-fold symmetry between transmembrane helices TM1–4 and TM5–8. Superposition of TM1–4 and TM5–8 ...
Crystal structures of ApUbiA and AfUbiA reveal the binding of 2 Mg2+ ions in complexes with GPP or GSPP; while there are differences in metal coordination polyhedra between structures, some of these differences may be attributable to the lower resolution (3.6 Å) of the ApUbiA structure128 in comparison with the higher resolution of the AfUbiA structure (2.4 Å).41 Two basic residues, an arginine and a lysine residue, also interact with the diphosphate group of GPP and presumably assist the Mg2+ ions in triggering substrate ionization.41 The reaction mechanism proposed on the basis of the higher resolution AfUbiA structure is shown in Figure Figure2323. This mechanism involves the characteristic ionization-condensation-elimination sequence of a class I terpenoid synthase, notable aspects of which include stabilization of the allylic carbocation intermediate through cation−π interactions with a highly conserved tyrosine residue and stereospecific proton elimination mediated by the inorganic pyrophosphate coproduct.41
Proposed mechanism of AfUbiA, a member of the UbiA superfamily that adopts the α fold of a class I terpenoid synthase. Reprinted from ref (41). Creative Commons Attribution 4.0 International Public License, http://creativecommons.org/licenses/by/4.0/legalcode ...
Perhaps the simplest reaction catalyzed by a terpenoid synthase is the conjugate 1,4-elimination reaction with DMAPP, yielding inorganic pyrophosphate and the volatile C5 hydrocarbon isoprene (Figure Figure2424). This reaction has tremendous ecological importance. Isoprene emission by plant life was first discovered more than 60 years ago by Sanadze,129 and worldwide atmospheric isoprene emissions are approximately 100 billion kg/year.130 The magnitude of hydrocarbon production by plant life is sometimes not fully appreciated, especially as it exceeds that generated by human activity.131−133 Rayleigh or Tyndall scattering of sunlight from atmospheric isoprene gives rise to the eponymous blue haze observed in the Blue Ridge Mountains even on otherwise clear days; such blue hazes were also described in the notebooks of Leonardo Da Vinci,134 who incorporated this feature into artwork such as the Mona Lisa (Figure Figure2525) and the Madonna of the Yarnwinder (Lansdowne Madonna).
1,4-conjugate elimination reaction of DMAPP yields isoprene as catalyzed by isoprene synthase. Similar reactions are catalyzed by myrcene synthase and farnesene synthase using substrates GPP and FPP, respectively. OPP = diphosphate, HOPP = inorganic pyrophosphate. ...
Mona Lisa. Leonardo da Vinci, ca. 1503. Oil on wood panel. Although somewhat yellowed, the faint blue haze in the background was inspired by the isoprene-derived blue haze in the hills of the Tuscan countryside, as recorded in Leonardo’s notebooks. For ...
Isoprene is also generated by certain bacteria and is generated at particularly high levels by Bacillus subtilis.135 Additionally, isoprene is the principal hydrocarbon identified in human breath;136 exhaled isoprene concentrations in a crowded football stadium are sufficiently high that they are measured to spike when goals are scored or when other exciting events occur on the field of play that inspire the roar of the crowd.137 Isoprene in human breath is also being studied as a biomarker for high cholesterol levels,138−140 muscular dystrophy,141 fibrotic liver disease,142 diabetes,143 and lung cancer.144 In bacteria and plants, isoprene is generated by isoprene synthase, a hemiterpene synthase.145−147 No human isoprene synthase enzyme is known, but nonenzymatic solvolysis of DMAPP may account for some,138 but not all,145 isoprene generation. It is conceivable that the 1,4-conjugate elimination reaction of DMAPP is catalyzed to some degree by another terpenoid biosynthetic enzyme in vivo, particularly given the correlation between isoprene and cholesterol levels.138−140
Plant isoprene synthase is a Mg2+-dependent class I terpenoid synthase.148−150 The enzyme from the gray poplar hybrid Populus × canescens has been cloned and expressed in E. coli.151 The crystal structure of this enzyme reveals a dimer of 64-kD subunits, each of which exhibits αβ domain architecture (Figure Figure2626).152 Dimeric quaternary structure enables cooperativity in catalysis, since native isoprene synthase exhibits a sigmoidal kinetic profile with a Hill coefficient of 1.9.152 The crystal structure of the complex with the unreactive DMAPP analogue dimethylallyl-S-thiolodiphosphate reveals that three Mg2+ ions are coordinated by metal-binding motifs on helices D and H and the substrate diphosphate group.152 The first metal-binding motif on helix D, D345DXXD, coordinates to Mg2+A and Mg2+C and is similar to the aspartate-rich motif first identified in prenyltransferases; the second metal-binding motif on helix H has a divergent amino acid sequence, N489DXXSXXXE, which chelates Mg2+B (metal ligands are in boldface) and is referred to as the NSE motif. While two aspartate-rich DDXXD metal-binding motifs almost always signal a prenyltransferase, the combination of one DDXXD motif and one NSE motif generally signals a lyase or cyclase enzyme. The substrate binding cleft of isoprene synthase is hydrophobic and relatively shallow to confer specificity for the C5 substrate, and the base of the cleft is largely defined by the aromatic side chains of F338 and F485 (Figure Figure2727).
Stereoview showing that each isoprene synthase monomer adopts αβ domain architecture, in which the α domain (blue) contains the characteristic aspartate-rich (red) and NSE (orange) metal-binding motifs. The substrate analogue dimethylallyl- ...
Stereoview showing a cut-away view of the active site surface of isoprene synthase complexed with the unreactive substrate analogue dimethylallyl-S-thiolodiphosphate (salmon) superimposed on the crystal structure of the monoterpene cyclase bornyl diphosphate synthase ...
The conformation of the bound substrate analogue in the active site of isoprene synthase suggests that the diphosphate group of DMAPP is ideally located to serve as the general base to deprotonate the allylic cation resulting from metal-triggered DMAPP ionization.152 While the α-phosphate group of the diphosphate group is closest to the (Z)-methyl group of the allylic cation, potentially indicating regiospecific proton elimination to generate isoprene, recent studies using isotopically labeled DMAPP153 indicate that this elimination reaction is not regiospecific (Figure Figure2828).154
If the proton elimination step of isoprene synthase were regiospecific then (E)-[4,4,4-2H3]DMAPP would exclusively yield [4,4,4-2H3]isoprene and (Z)-[4,4,4-2H3]DMAPP would exclusively yield [1,1-2H2]isoprene. However, each isotopically labeled substrate ...
Isoprene synthase has become very important in recent years in biotechnology, since isoprene is a critical starting material for the generation of synthetic rubber for tire manufacturing. As the demand for synthetic rubber has increased in recent years, so too has the demand for isoprene increased. Since commercial feedstocks of isoprene derive mainly from petrochemical refinement, alternative sources of isoprene are being explored through metabolic engineering in bacteria and yeast to develop robust and renewable sources of isoprene.155−159 Such “bioisoprene” was used to manufacture the world’s first “green” rubber tire through a business collaboration between Goodyear and Genencor,160 presented at the 2009 United Nations Climate Change Conference in Copenhagen, Denmark.
The 1,4-conjugate elimination reaction also occurs with substrates GPP and FPP (Figure Figure2424; other regiomeric elimination isomers can be formed as well), and these reactions are catalyzed in biological systems by myrcene synthase and (E)-β-farnesene synthase, respectively.161−165 Although crystal structures of these class I terpenoid synthases have not yet been reported, a separate farnesene synthase active site is found “moonlighting” in the structure of albaflavenone synthase (but this bifunctional enzyme does not adopt the canonical α fold of a class I terpenoid synthase).166 These enzymes are believed to operate through a mechanism similar to that of isoprene synthase, each proceeding through an allylic carbocation intermediate resulting from metal-triggered ionization of the diphosphate moiety. In the (E)-β-farnesene synthase reaction, Allemann and colleagues demonstrate that inorganic pyrophosphate can recombine with the initially formed allylic cation to form a trans-nerolidyl diphosphate intermediate, which then reionizes to complete the elimination reaction.154 Notably, farnesene synthase is currently utilized in a large-scale metabolic engineering process to generate farnesene-based jet fuel.167−169
Class I terpenoid cyclases are metal-dependent enzymes that catalyze the cyclization of isoprenoid diphosphate substrates to form hydrocarbon products containing one or more fused rings and stereocenters. Class I terpenoid cyclases adopt the same α fold as that first observed in the crystal structure of avian FPP synthase.58 These enzymes are found with α, αβ, and αβγ domain architectures, and bifunctional enzymes are found with αα (class I-class I) and αβγ (class I-class II) domain architectures. Like FPP synthase, these cyclases contain metal-binding motifs on helices D and H on the upper walls of their active sites. However, while the first metal-binding motif on helix D is typically an aspartate-rich DDXXD motif, the second metal-binding motif on helix H is almost always (N,D)D(L,I,V)X(S,T)XXXE in a cyclase (boldface residues indicate metal ligands).170 The second metal-binding motif is accordingly designated the NSE or DTE motif. As noted in section 4, isoprene synthase has DDXXD and NSE metal-binding motifs, so the presence of an NSE/DTE motif usually distinguishes isoprenoid diphosphate lyases and terpenoid cyclases from prenyltransferases containing two DDXXD metal-binding motifs such as FPP synthase (section 2). The remainder of this section outlines structure-mechanism relationships for class I terpenoid cyclases, focusing mainly on cyclases with known crystal structures that utilize GPP, FPP, and GGPP substrates.
The C10 isoprenoid GPP (Figure Figure11) is the minimum-length cyclization substrate in terpenoid biosynthesis, and the monoterpene cyclization cascade typically proceeds through the α-terpinyl carbocation resulting from C1–C6 bond formation (Figure Figure2929). A structurally and stereochemically diverse array of carbon skeletons result from subsequent reactions of this important intermediate as governed by its conformation, which in turn is enforced by the three-dimensional contour of the enzyme active site. The active site contour serves as a template for catalysis, ensuring that the substrate and intermediates adopt only those conformations leading to the formation of the correct product(s). Accordingly, active site contours are productlike, especially for high-fidelity cyclases, to ensure the generation of a specific product. Additionally notable is the role of water in the chemistry of GPP cyclization: the final carbocation intermediate in some monoterpene cyclization cascades is quenched by the addition of a solvent molecule, so this solvent molecule must be strictly controlled so that it does not prematurely quench carbocation intermediates in the cyclization cascade. Various aspects of these features are evident in structural and functional studies of monoterpene cyclases that have yielded crystal structures to date.
Isolated from Salvia officinalis (culinary sage), (+)-bornyl diphosphate synthase is a homodimer of 64-kD subunits that catalyzes an unusual reaction sequence, in that the substrate diphosphate group dissociates and rebinds to carbocation intermediates twice during the GPP cyclization cascade (Figure Figure3030).171 Molecular details of the catalytic mechanism have been delineated in a variety of enzymological studies,172−175 including positional isotope exchange experiments conclusively demonstrating that the same diphosphate ester oxygen atom makes the prenyl linkage of substrate GPP and product (+)-bornyl diphosphate.176,177 Formation of the 6-membered ring of the α-terpinyl cation intermediate first requires isomerization of the C2–C3 double bond from a trans configuration to a cis configuration. This isomerization cannot occur directly, of course, due to the ~60 kcal/mol rotational barrier of the double bond. As summarized in Figure Figure3030, this isomerization is achieved by metal-triggered ionization of GPP to form a trans-allylic cation intermediate, followed by readdition of inorganic pyrophosphate at C3 to yield linalyl diphosphate. Only when the C2–C3 single bond is fully formed can rotation occur from a transoid to a cisoid conformation which, following reionization to yield a cis-allylic cation intermediate, enables C1–C6 bond formation to generate the α-terpinyl cation. The conformation of this cation is held in such a way that a second carbon–carbon bond is formed to yield the 2-bornyl cation, which is ultimately quenched by readdition of inorganic pyrophosphate to yield (+)-bornyl diphosphate.
GPP cyclization cascade catalyzed by (+)-bornyl diphosphate synthase. Positional isotope exchange experiments show that the prenyl diphosphate ester oxygen atom of the substrate (red) is the same as that of the product. Aza analogues of carbocation intermediates ...
The crystal structures of (+)-bornyl diphosphate synthase and its complexes determined at resolutions of 2.0–2.4 Å were the first of a monoterpene cyclase and revealed αβ domain architecture23 similar to that first observed in epi-aristolochene synthase.37 Such αβ domain architecture is characteristic for plant terpenoid cyclases. Two monomers of (+)-bornyl diphosphate synthase assemble in antiparallel fashion with an extensive dimer interface mediated by their α domains (Figure Figure3131). Although this monoterpene cyclase is not an allosteric enzyme, it is interesting that its quaternary structure is identical to that of isoprene synthase, which exhibits cooperativity in the 1,4-conjugate elimination reaction with DMAPP (section 4). It is further interesting to note that this quaternary structure is also shared with other dimeric plant monoterpene cyclases such as (−)-limonene synthase (section 5.1.2) (Figure Figure3131). More recently, identical quaternary structures have been observed for the αβ monoterpene cyclases (+)-limonene synthase (section 5.1.3) and γ-terpinene synthase (section 5.1.4), so the common quaternary structures illustrated in Figure Figure3131 may represent a universal dimerization mode for plant terpenoid cyclases with αβ domain architecture.
Isoprene synthase (A), (+)-bornyl diphosphate synthase (B), and (−)-limonene synthase adopt similar βα:αβ dimeric quaternary structures with more than 1000 Å2 buried surface area at the dimer interface. ...
Crystal structures of several complexes of (+)-bornyl diphosphate synthase complexed with inorganic pyrophosphate, aza analogues of carbocation intermediates (illustrated in Figure Figure3030), and the product (+)-bornyl diphosphate show that 3 Mg2+ ions liganded by DDXXD and DTE motifs hold a diphosphate group rigidly in place, effectively capping the mouth of the active site cavity (Figure Figure3232).23 Assisting the metal ions in diphosphate complexation are three basic residues, R314, R493, and K512. Thus, 3 metal ions and 3 basic residues comprise a diphosphate recognition motif in the active site of this and other terpenoid cyclases.66 Comparison of the unliganded and liganded enzyme structures reveals that the binding of 3 metal ions and the diphosphate group triggers a conformational change from an open to a closed active site conformation (Figure Figure3333). The second aspartate in the DDXXD motif, D352, hydrogen bonds with R314 to stabilize the closed active site conformation; since R314 also donates a hydrogen bond to the diphosphate group, this interaction directly links the binding of the substrate diphosphate group with active site closure. The conformational change from an open active site conformation to a closed active site conformation ensures that highly reactive carbocation intermediates in catalysis are protected from bulk solvent, and such a conformational change is common to all class I terpenoid synthases. Similar conformational changes are observed in FPP synthase upon ligand binding in the active site (section 2.1).
Three metal ions and three basic residues (R314, R493, and K512) are critical for the molecular recognition of the diphosphate group in the active site of (+)-bornyl diphosphate synthase. This is a common theme for diphosphate recognition in the active ...
(A) Stereoview of unliganded (+)-bornyl diphosphate synthase, looking into the active site in the α domain (blue). Disordered polypeptide segments are indicated by dotted lines and include the N-terminal segment of the β domain (green). ...
The active site of (+)-bornyl diphosphate synthase serves as a template that strictly enforces the left-handed helical conformation of GPP required for the generation of (+)-bornyl diphosphate via the (4R)-α-terpinyl cation. However, the template is not completely perfect, and minor quantities of alternative cyclic monoterpenes such as pinenes, camphene, and limonene are also generated.171 Interestingly, F578 and W323 form part of the active site contour, and they may also stabilize carbocation intermediates through cation−π interactions. Two or three aromatic residues are typically found in most terpenoid cyclase active sites where they may serve this function, stabilizing high-energy carbocation intermediates as well as the transition states flanking these intermediates.
Surprisingly, a water molecule (no. 110) is trapped in the active site of (+)-bornyl diphosphate synthase in the crystal structures of all enzyme-ligand complexes, including the complex with product (+)-bornyl diphosphate itself.23 This water molecule forms hydrogen bonds with the diphosphate group of the ligand, Y426, and the backbone carbonyl of S451. This water molecule, located near the geminal-dimethyl group of (+)-bornyl diphosphate (Figure Figure3434), likely remains bound in the active site upon binding of the actual substrate GPP and is presumably restrained in place so that it cannot prematurely quench carbocation intermediates in the cyclization cascade. Thus, a trapped water molecule can serve as part of the active site contour in a terpenoid cyclase, contributing to the conformational control of substrate binding. In terpenoid cyclases that generate hydroxylated products, the precise control of a trapped water molecule is critically important for the fidelity of the cyclization cascade in generating the desired product.
Stereoview of the simulated annealing electron density map of the (+)-bornyl diphosphate synthase-Mg2+3-(+)-bornyl diphosphate complex. Metal coordination and hydrogen bond interactions are indicated by solid and dashed black lines, respectively. Water molecule ...
Curiously, some aza analogues of carbocation intermediates illustrated in Figure Figure3030 do not bind in catalytically productive conformations. For example, 3-aza-2,3-dihydrogeranyl diphosphate does not bind with the left-handed helical conformation that would be required for GPP to generate the (4R)-α-terpinyl cation intermediate. This phenomenon is not uncommon for the binding of flexible unreactive substrate analogues in terpenoid cyclase active sites and may result from the imperfect mimicry between the analogue and the substrate or intermediate after which it is designed. Moreover, the thermodynamically stable structure corresponding to a crystalline enzyme-analogue complex at equilibrium does not necessarily correspond to the structure of the enzyme-carbocation intermediate complex with the lowest accessible transition state. In other words, achieving productive conformations along a reaction pathway during catalysis is under kinetic control and not thermodynamic control. This distinction may account for anomalous binding modes observed for certain unreactive analogues of natural carbocation intermediates.178
Notably, however, other analogues bind in a productive fashion. For example, the (+)-bornyl diphosphate synthase-Mg2+3-2-azabornane complex aligns nearly perfectly with that of the complex with 3 Mg2+ ions and the actual product, (+)-bornyl diphosphate, comparison of these two structures serves as a “before and after” picture of the C–O bond-forming reaction with the 2-bornyl cation (Figure Figure3535).
Stereoview of the superposition of the (+)-bornyl diphosphate synthase complexes with Mg2+3 and inorganic pyrophosphate (yellow), inorganic pyrophosphate and 2-azabornane (magenta), and (+)-bornyl diphosphate (blue). The molecular recognition of the diphosphate moiety ...
The reaction mechanism of (+)-bornyl diphosphate synthase has been studied extensively using theoretical and computational methods.179−182 Notably, these studies indicate that the 2-bornyl cation represents a transition state, rather than an actual intermediate, that follows the formation of an intermediate pinyl cation; moreover, Major and Weitman show that the 2-bornyl cation is a bifurcation point on the free energy surface that can lead to formation of product (+)-bornyl diphosphate or side product camphene via the camphyl cation (Figure Figure3636).181 This sort of bifurcation was first observed in theoretical and computational studies of terpenoid biosynthesis in studies of diterpene synthase mechanisms by Hong and Tantillo.183 In (+)-bornyl diphosphate synthase, a bifurcated free energy surface is one factor that can compromise cyclization fidelity by enabling the formation of alternate side products, as long as the conformations required to generate them can be accommodated by the active site template. If the 2-bornyl cation represents a transition state rather than an intermediate in the catalytic mechanism, then 2-azabornane is a productlike transition state analogue.
Theoretical and computational chemistry studies indicate that the 2-bornyl cation is actually a transition state at a bifurcation point on the reaction coordinate leading to either (+)-bornyl diphosphate or the camphyl cation, which can undergo proton ...
The generation of (−)-limonene is perhaps the simplest cyclization reaction catalyzed by a class I terpenoid cyclase, since the α-terpinyl carbocation intermediate merely undergoes an E1 elimination reaction (Figure Figure3737). Notably, the terpinyl carbocation in this reaction is the enantiomer of that formed in the (+)-bornyl diphosphate synthase reaction (Figure Figure3030), indicating that the active site template of (−)-limonene synthase enforces a right-handed binding conformation for substrate GPP. (−)-Limonene synthase from Mentha spicata(184−188) is a high-fidelity enzyme and generates 94% (−)-limonene; approximately 4% pinene isomers and 2% myrcene side products are also generated, the latter of which reflects unsuccessful C1–C6 bond formation.189
The crystal structure of (−)-limonene synthase determined at 2.7 Å resolution reveals a tertiary structure with αβ domain architecture and dimeric βα:αβ quaternary structure identical to that observed for isoprene synthase and (+)-bornyl diphosphate synthase (Figure Figure3131).190 The β domain has no known catalytic function, so it is probably an evolutionary vestige; however, the N-terminal polypeptide segment can assist in stabilizing the closed active site conformation, as observed in the crystal structure of (+)-bornyl diphosphate synthase.23 Intron conservation patterns in plant terpenoid cyclases suggest that αβ cyclases evolved from ancestral αβγ cyclases in which both domains were catalytically active.191 A modern-day example of such a bifunctional αβγ cyclase is abietadiene synthase, which catalyzes an overall reaction sequence involving an initial class II cyclization reaction followed by a class I cyclization reaction (section 7.2).45,192
Surprisingly, cocrystallization of (−)-limonene synthase with the substrate analogue 2-fluorogeranyl diphosphate yields the structure of the complex with 2-fluoro-(3S)-linalyl diphosphate.190 While fluorination of the substrate at C2 is intended to destabilize and thwart formation of the allylic cation that would be formed upon substrate ionization, Croteau and colleagues demonstrate that this substrate analogue is slowly reactive in solution and in the crystal.190 The direct cocrystallization of (−)-limonene synthase with racemic 2-fluorolinalyl diphosphate yields an electron density map that can be fit with either enantiomer in helical conformations required for cyclization to products of respective stereochemistry. Interestingly, cocrystallization of (−)-limonene synthase with 2-fluorogeranyl diphosphate yields the crystal structure of bound 2-fluorolinalyl diphosphate with a catalytically nonproductive conformation. The relatively low 2.7 Å resolution of these crystal structure determinations limits the interpretation of electron density maps (Figure Figure3838), but these maps nonetheless show how the binding conformations of 2-fluorolinalyl diphosphate differ depending on how the complexes were prepared. This observation may suggest that the substrate need not be in the productive helical conformation required for cyclization in the first step of the reaction, the isomerization of geranyl diphosphate to linalyl diphosphate. Indeed, an extended conformation may be required for this isomerization step, since a helical conformation could conceivably hinder the allylic diphosphate isomerization and transoid-cisoid conformational change preceding the cyclization step (two different substrate conformations are similarly implicated in the isomerization and cyclization steps of amorphadiene synthase193). The catalytically productive helical conformation can then be achieved after the isomerization step to complete the cyclization sequence.190
(a) Stereoview of an electron density map calculated with Fourier coefficients 2|Fo| – |Fc| and phases calculated from the final model showing (−)-limonene synthase complexed with substrate analogue 2-fluorolinalyl diphosphate generated enzymatically ...
Recently, the crystal structures of unliganded (+)-limonene synthase from navel orange (Citrus sinensis) and its complexes with fluorinated substrate analogues have been reported at resolutions of 2.2–2.4 Å.194,195 These structures reveal the characteristic αβ fold of a plant terpenoid cyclase with dimeric quaternary structure. However, even though 3 Mn2+ ions are bound in the complexes with fluorinated substrate analogues, the enzyme active site remains in the open conformation, apparently blocked from forming a fully closed conformation by crystal lattice contacts. Even so, comparison with the crystal structure of (−)-limonene synthase complexed with 2-fluorolinalyl diphosphate (45% amino acid sequence identity) provides an excellent opportunity for studying the active site closure mechanism. Structural changes triggered by the binding of substrate analogue and 3 metal ions are generally similar to those observed in (+)-bornyl diphosphate synthase (Figure Figure3333).23 Interestingly, however, the side chain of highly conserved residue Y565 on helix J of (−)-limonene synthase undergoes a conformational transition from an “out” conformation in the unliganded enzyme to an “in” conformation in the analogue-bound enzyme. The Y565F mutant exhibits a 50-fold loss of catalytic activity, but the molecular basis of Y565 function is not yet fully established. Oprian and colleagues suggest that it might help block water from entering the active site, stabilize reaction intermediates, and/or stabilize metal ion binding through hydrogen bonding to an aspartate metal ligand.194
The crystal structures of (+)-limonene synthase and (−)-limonene synthase are the first to be solved of terpenoid cyclases that generate product enantiomers. The generation of (+)-limonene requires a left-handed binding conformation for substrate GPP and intermediate linalyl diphosphate, whereas the generation of (−)-limonene requires a right-handed binding conformation for substrate GPP and intermediate linalyl diphosphate (Figure Figure3737). Despite the dramatically different substrate binding conformations required for catalysis, it is striking that the amino acid residues lining the active sites of (+)-limonene synthase and (−)-limonene synthase are highly conserved (Figure Figure3939).194 What, then, provides the stereochemical imperative for substrate binding and catalysis?
On the basis of analysis of enzyme complexes with fluorinated analogues, Oprian and colleagues suggest that active site residues M458/I450 and N345/I336 of (−)-limonene synthase/(+)-limonene synthase govern stereospecificity of the respective GPP cyclization cascades.195 The Sδ atom of the M458 side chain would sterically clash with the terminal methyl groups of 2-fluorolinalyl diphosphate if it adopted an incorrect left-handed conformation in the active site of (−)-limonene synthase, and the Cγ2 group of I336 would similarly clash with the terminal methyl groups of 2-fluorogeranyl diphosphate if it adopted an incorrect right-handed conformation in the active site of (+)-limonene synthase (Figure Figure3939). This is an elegant demonstration showing how the active site contour of a terpenoid cyclase is a template that directs structure and stereochemistry in a cyclization cascade.
The generation of γ-terpinene from GPP proceeds through the universal α-terpinyl cation (Figure Figure2929), which then undergoes a 1,2-hydride shift to yield the terpinen-4-yl cation, which after endocyclic proton elimination yields γ-terpinene (Figure Figure4040). This reaction is important in the biosynthetic pathways of carvacrol and thymol, antimicrobial monoterpenoid phenols responsible for the distinctive odors and flavors of oregano and thyme, respectively.196,197 To date, γ-terpinene synthase enzymes have been identified in several plant species.198
Cyclization reaction catalyzed by γ-terpinene synthase is the first committed step in the biosynthesis of carvacrol and thymol. This monoterpene cyclase adopts the characteristic αβ domain architecture of a plant monoterpene cyclase. ...
Recently, γ-terpinene synthase from Thymus vulgaris (common thyme) has been cloned and expressed, and its crystal structure has been determined in the unliganded state at 1.65 Å resolution.199 The crystal structure reveals the characteristic αβ domain architecture (Figure Figure4040) and βα:αβ dimeric quaternary structure of plant monoterpene and hemiterpene synthases. Although crystal structures of γ-terpinene synthase complexed with substrate analogues are not yet available, the structure of the unliganded enzyme will serve as a valuable reference point for understanding structure–activity relationships in site-specific mutants.
Cineole, also known as 1,8-cineole or more popularly as eucalyptol, is the primary component of eucalyptus oil and has a pleasant spicy aroma and taste. Eucalyptol is also being explored as a potential therapeutic agent due to its anti-inflammatory and antioxidant properties.200,201 Although cineole derives from the universal cyclic monoterpene intermediate α-terpinyl cation (Figure Figure2929), this cyclic monoterpene is unique in that the final product of the overall cyclization cascade is an ether. Strictly speaking, two cyclization sequences are catalyzed in the active site of cineole synthase: (1) a canonical metal-triggered class I GPP cyclization reaction leading to formation of the α-terpinyl cation, quenched by a water molecule trapped in the active site to form α-terpineol and (2) a protonation-induced cyclization reaction, in which the carbon–carbon double bond of α-terpineol is protonated to generate a tertiary carbocation intermediate which subsequently undergoes intramolecular attack by the hydroxyl group to yield cineol (Figure Figure4141).
Reaction sequence catalyzed by the monoterpene cyclase cineole synthase. The ionization-dependent class I cyclization reaction of GPP forming α-terpineol is followed by a second protonation-induced cyclization reaction to yield 1,8-cineole.
The crystal structure of cineole synthase from Salvia fruticosa (Greek oregano) determined at 1.95 Å resolution in the unliganded state reveals characteristic αβ domain architecture (Figure Figure4242) and βα:αβ dimeric quaternary structure of plant monoterpene and hemiterpene synthases.202 Although the crystal structure of cineole synthase provides limited mechanistic inferences in the absence of bound substrate analogues, superposition with (+)-bornyl diphosphate synthase complexed with the analogue 3-aza-2,3-dihydrogeranyl diphosphate23 confirms the location of the active site (Figure Figure4242). The authors suggest that a water molecule hydrogen bonded to N338 in cineole synthase is trapped in the enzyme active site along with GPP to facilitate hydroxylation of the α-terpinyl cation and formation of the α-terpineol intermediate.202 However, the trapped water molecule hydrogen bonded to Y426 and inorganic pyrophosphate in (+)-bornyl diphosphate synthase (Figure Figure3434) could similarly be trapped and hydrogen bonded to conserved residue Y420 and inorganic pyrophosphate in the active site of cineole synthase. Since this water molecule is located near the geminal dimethyl group of 2-azabornane and (+)-bornyl diphosphate, which accordingly would be in the vicinity of the geminal dimethyl group of the α-terpinyl cation intermediate, it is possible that this water molecule hydroxylates the α-terpinyl cation in the cineole synthase mechanism. Given the likely role of coproduct inorganic pyrophosphate as a potential general base-general acid catalyst,80 inorganic pyrophosphate may deprotonate this water molecule upon hydroxylation of the α-terpinyl cation and then protonate the carbon–carbon double bond of α-terpineol to enable cineole formation as shown in Figure Figure4141.
Structure of cineole synthase; the α domain is green and the β domain is blue. The inset shows a close-up view of the active site, with the bound conformation of 3-aza-2,3-dihydrogeranyl diphosphate modeled in based on the corresponding ...
Kampranis and colleagues use the cineole synthase structure to guide intriguing protein engineering experiments.202 The N338I substitution results in the generation of 48% sabinene and 37% limonene as major products, with no generation of hydroxylated products. The N338A substitution enlarges the active site and converts cineole synthase into a sesquiterpene synthase capable of utilizing FPP as a substrate, generating 49% trans-α-bergamotene as its major cyclic product. Thus, N338 appears to be a “hot spot” for the evolution of cyclization specificity and fidelity in this terpenoid cyclase.
While the great majority of terpenoid cyclases utilize canonical C5n isoprenoid diphosphate substrates such as GPP, there are occasional examples of cyclases capable of utilizing modified substrates in the laboratory or in nature. A rare example found in nature is methylisoborneol synthase from Streptomyces coelicolor A3(2), a monoterpene synthase that catalyzes the cyclization of a noncanonical C11 substrate to form 2-methylisoborneol (Figure Figure4343). This unusual terpenoid is a potent odorant and is responsible for the musty smell of contaminated water,203−205 but it is also responsible for the pleasing, earthy aroma of Brie and Camembert cheeses.206 The novel cyclization substrate, 2-methyl-GPP, derives from the S-adenosylmethionine-dependent methylation of GPP in a reaction catalyzed by geranyl diphosphate methyltransferase.207−211
Reaction sequence catalyzed by monoterpene synthase methylisoborneol synthase, which utilizes the novel C11 substrate 2-methylgeranyl diphosphate. Isomerization to (3R)-2-methyllinalyl diphosphate does not require the substrate to be in a cyclization-competent conformation. ...
Methylisoborneol synthase is the first example of a naturally occurring terpenoid cyclase that utilizes a noncanonical acyclic substrate,209,210 and several orthologues are found in other bacterial and cyanobacterial species. The crystal structures of the unliganded enzyme and its complexes with the substrate analogues geranyl-S-thiolodiphosphate, 2-fluorogeranyl diphosphate, and 2-fluoroneryl diphosphate determined at resolutions of 1.8–2.0 Å reveal several unusual features.212,213 First, although most bacterial cyclases consist of a single α domain, methylisoborneol synthase contains an additional 13-kD N-terminal domain with more than 25% proline content; surprisingly, this domain is completely disordered in the crystal. SDS–PAGE analysis confirms that this domain is present in the protein construct used for crystallization. The function of this proline-rich domain is unknown, but it is conserved in most methylisoborneol synthase orthologues; it is not found in any other enzyme.
Methylisoborneol synthase forms a dimer in which enzyme active sites are oriented in parallel fashion, similar to the parallel quaternary structure of FPP synthase but contrasting with the antiparallel quaternary structure of plant monoterpene cyclases such as (+)-bornyl diphosphate synthase (Figure Figure4444). Even so, the same face of the α domain is utilized for dimerization in bacterial and plant terpenoid synthases, although different helices on this face of the protein can be involved. The crystal structures of complexes between methylisoborneol synthase and different substrate analogues reflect incomplete binding of metal ions by the aspartate-rich and NSE motifs on helices D and H, respectively; only Mg2+B and Mg2+C accommodate the diphosphate group of the substrate analogue, suggesting incomplete active site closure. Additionally, the enzyme must sequester a water molecule in the active site for use in quenching the final carbocation intermediate, but no clues are evident in the active site structure as to where such a water molecule might reside during catalysis so as to not prematurely quench the cyclization cascade. A stereoview of the enzyme complex with 2-fluorogeranyl diphosphate is shown in Figure Figure4545A, and electron density maps showing the binding of this analogue as well as geranyl-S-thiolodiphosphate are found in Figure Figure4545 (panels B and C, respectively).
Dimeric quaternary structure of bacterial methylisoborneol synthase (MIBS), avian farnesyl diphosphate synthase (FPPS), and (+)-bornyl diphosphate synthase (BPPS). Catalytic α domains are blue, and the β domains of (+)-bornyl diphosphate ...
(A) Stereoview of the complex between methylisoborneol synthase and 2-fluorogeranyl diphosphate. The aspartate-rich and NSE metal-binding motifs are red and orange, respectively. (B) Simulated annealing omit map showing the binding of 2-fluorogeranyl ...
Surprisingly, cocrystallization of methylisoborneol synthase with the fluorinated analogue of the tertiary allylic diphosphate intermediate, 2-fluorolinalyl diphosphate, yields the crystal structure of the complex with the primary cis-allylic diphosphate, 2-fluoroneryl diphosphate (Figure Figure4646).213 This unexpected reactivity indicates that fluorination does not completely block ionization of the diphosphate group or reaction of the resultantly formed allylic cation. 2-Fluorolinalool is the major product isolated after incubation of methylisoborneol synthase with 2-fluorolinalyl diphosphate for 12 h; however, minor amounts of camphor are also detected, indicating that ionization and cyclization of the fluorinated substrate analogue can slowly proceed to yield 2-fluoroisoborneol, which rapidly eliminates HF to yield camphor.
Stereoview of the simulated annealing omit map of 2-fluoroneryl diphosphate (2FNPP) bound in the active site of methylisoborneol synthase. Only two metal ions bind in this complex; metal coordination interactions are indicated by red dashed lines. This ...
In contrast with GPP cyclization reactions, which typically proceed through an ionization-recombination-reionization sequence to enable C1–C6 bond formation and generation of the α-terpinyl cation intermediate (Figure Figure2929), FPP cyclization reactions can proceed through C1–C6, C1–C7, C1–C10, and C1–C11 bond forming reactions depending on which carbon–carbon double bond reacts with the initially formed allylic carbocation (Figure Figure4747).214 In turn, the resulting carbocation intermediate can undergo further cyclization reactions, hydride transfers, methyl migrations, etc., before termination of the reaction sequence. Thus, in comparison with monoterpene cyclization reactions, increasingly diverse arrays of carbon skeletons result from sesquiterpene cyclization reactions owing to the additional carbon–carbon bond-forming trajectories available to the larger substrate.
Possible trajectories of initial carbon–carbon bond formation in sesquiterpene cyclization reactions (OPP = diphosphate).
The active site contour in the α domain of an enzyme with overall α or αβ domain architecture serves as a template for catalysis by a sesquiterpene cyclase, ensuring that substrate FPP and subsequently formed intermediates adopt only those conformations leading to formation of the correct product(s). Accordingly, active site contours are productlike, especially for high-fidelity cyclases, to ensure the generation of a specific product. Numerous sesquiterpene cyclases have yielded crystal structures in the past two decades, and each structure has advanced our understanding of the general features underlying the chemistry of terpenoid cyclization reactions in biology. Below, cyclases are discussed in order of complexity of their initial cyclization reactions, and then in order of increasing complexity of their overall cyclization cascades: C1–C6 cyclases (α-bisabolol synthase, α-bisabolene synthase, epi-isozizaene synthase, trichodiene synthase); C1–C10 cyclases (selinadiene synthase, germacradien-4-ol synthase, aristolochene synthase, epi-aristolochene synthase, hedycaryol synthase, (+)-δ-cadinene synthase); and a C1–C11 cyclase (pentalenene synthase).
The sesquiterpene alcohol α-bisabolol is responsible for many of the beneficial effects of chamomile (Matricaria recutita), a common herbal remedy.215,216 α-Bisabolol is reported to exhibit antibacterial and anti-inflammatory properties,217−219 and its derivatives are also potential cancer chemotherapeutic agents through their activity in blocking the expression of the serine/threonine kinase AKT.220−222 Interestingly, α-bisabolol is also a biosynthetic precursor of hernandulcin, a natural sweetener generated in the leaves of Lippia dulcis. Also known as Phyla dulcis, this plant is a perennial herb indigenous to Central America where it is also known as Aztec sweet herb or honey herb.223
The generation of α-bisabolol from FPP is catalyzed by α-bisabolol synthase. This reaction requires an ionization-recombination-reionization sequence224 identical to that described in the previous section leading to formation of the α-terpinyl cation, so as to ensure C1–C6 bond formation and generation of the 6-membered ring (Figure Figure4848).225 The resulting bisabolyl cation is then quenched by a solvent molecule that is trapped in the active site along with the substrate. This solvent molecule must be suitably restrained so as not to prematurely quench the reaction sequence prior to formation of the bisabolyl carbocation intermediate.
Mechanisms of FPP cyclization catalyzed by α-bisabolol synthase (AaBOS) and the penta-substituted mutant AaBOS-M2. These reaction sequences are compared with the reactions catalyzed by the related sesquiterpene cyclase amorphadiene synthase (AaADS). ...
α-Bisabolol synthase from Artemisia annua was recently cloned and expressed in E. coli, and the crystal structure of the unliganded wild-type enzyme has been determined at 2.0 Å resolution.225 The enzyme exhibits the characteristic αβ domain architecture of a plant terpenoid cyclase, with the nonpolar active site located exclusively in the α domain (Figure Figure4949). Interestingly, although the aspartate-rich metal-binding motif DDXXD is located on helix D as expected based on structural homology with other terpenoid cyclases, the NSE/DTE metal-binding motif on helix H appears as NGE; moreover, this segment is partially disordered, probably due to the lack of bound metal ions and ligand. The substitution of metal ligand serine with nonliganding residue glycine in this motif is occasionally observed in plant terpenoid cyclases, so the central serine or threonine residue in the NSE/DTE metal binding motif is occasionally dispensable for Mg2+B binding.226
Crystal structure of α-bisabolol synthase reveals αβ domain architecture. The active site resides in the α domain (catalytic domain) and the β domain has no known catalytic function. Reprinted with permission from ...
α-Bisabolol synthase is a relatively high-fidelity enzyme, in that the wild-type enzyme generates 92.8% α-bisabolol [minor side products include trans-α-bisabolene (1.1%), cis-α-bisabolene (1.2%), cis-γ-bisabolene (2.4%), β-bisabolene (1.9%), and β-sesquiphellandrene (0.51%)].225 Analysis of the active site structure led to the design and preparation of a penta-substituted mutant (V373N, L381A, I395V, N398I, and L399T) designated M2-α-bisabolol synthase. The major cyclization product generated by this mutant is γ-humulene (68.8%), indicating that the array of mutations enables anti-Markovnikov C1–C11 bond formation after the ionization-recombination-reionization sequence (Figure Figure4848). The 2.0 Å-resolution crystal structure of M2-α-bisabolol synthase225 reveals that the 5 amino acid substitutions cause minor changes in the conformations of helices G1 and F, but significant structural changes result in the three-dimensional shape of the active site pocket. These changes must influence the binding conformation of FPP and cause it to reorient so as to favor C1–C11 bond formation. These structural and functional studies illuminate structure-mechanism relationships in the related sesquiterpene cyclase amorphadiene synthase, which catalyzes the first committed step of artemisinin biosynthesis (Figure Figure4848).225
In response to insect attack, various conifer species generate complex terpenoid mixtures to ward off the attacking species.227−229 This chemical defense system consists of numerous volatile monoterpenes and sesquiterpenes, a mixture known informally as turpentine, as well as nonvolatile diterpene resin acids (rosin). The volatile terpenoids generated are typically toxic for the invading insect; these lower molecular weight species also solubilize the higher molecular weight diterpene resin acids, which plug the lesion caused by the insect bite after evaporation of the volatile components.230
The sesquiterpene mixture generated by conifer species in response to insect attack is particularly notable with regard to both short-term and long-term chemical defense strategies. Not only are these compounds immediately toxic when encountered by the invading insect, but they can also disrupt the insect life cycle, thereby comprising a second line of defense for long-term protection.231−234 For example, todomatuic acid and its methyl ester juvabionine inhibit the ability of insects to molt into adults, presumably due to mimicry of insect juvenile hormones.235 Todomatuic acid is believed to derive from α-bisabolene (Figure Figure5050A), and α-bisabolene synthase is accordingly upregulated in response to stem wounding in Abies grandis (grand fir).236
(A) The catalytic mechanism of α-bisabolene synthase proceeds through a characteristic ionization-recombination-reionization sequence in order to facilitate C1–C6 bond formation. The cyclization product, α-bisabolene, is a biosynthetic ...
α-Bisabolene synthase from Abies grandis has been cloned and expressed in E. coli, and its functional properties have been described.236 Using FPP as a substrate, α-bisabolene synthase generates >99% α-bisabolene, so it is a high-fidelity cyclase. Interestingly, the enzyme also converts GPP into (+)-limonene, but this activity is not believed to be biologically significant due to the localization of α-bisabolene synthase in the cell cytosol (in plants, GPP and monoterpene cyclization reactions are compartmentalized in the plastid). The reaction sequence for α-bisabolene synthase (Figure Figure5050A) is analogous to that required for the generation of limonene (Figure Figure3737), in that both require the ionization-recombination-reionization sequence through a tertiary diphosphate intermediate to enable C1–C6 bond formation.
The 2.14 Å resolution crystal structure of unliganded α-bisabolene synthase46 reveals αβγ domain architecture (Figure Figure5050B), in which the class I active site resides in the α domain; the βγ domains have no known catalytic function, other than perhaps for the N-terminal polypeptide segment to help cap the active site upon substrate binding. Crystal structures of complexes with substrate analogues as well as bisphosphonate inhibitors such as alendronate, pamidronate, and etidronate reveal characteristic molecular recognition of the diphosphate moiety by 3 Mg2+ ions complexed by the aspartate-rich and DTE metal-binding motifs in the active site. The catalytic efficiency of the enzyme is relatively modest,46 with kcat/KM = 38 M–1 s–1.
Notably, α-bisabolene synthase has been utilized in metabolic engineering experiments in the development of new approaches for the generation of advanced biofuels in yeast and cyanobacteria.237,238 The chemical hydrogenation of α-bisabolene yields the saturated hydrocarbon bisabolane, which has physical and chemical properties quite similar to those of D2 diesel fuel.237 Engineering and economic calculations by Keasling and colleagues indicate that the commercial production of α-bisabolene through metabolic engineering would yield a product for approximately $6/gallon, which is currently more expensive than petrochemically derived D2 diesel fuel. Furthermore, large-scale hydrogenation of bioderived α-bisabolene would add to this cost. However, Keasling and colleagues anticipate terpene reduction reactions engineered in synthetic biology host systems using designer reductases, which could reduce the cost-per-gallon for a consumer-grade liquid fuel.237
Albaflavenone is an antibiotic generated in the Gram-positive soil bacterium Streptomyces coelicolor A3(2) that derives from the cytochrome P450 170A1-mediated oxidation of the tricyclic sesquiterpene epi-isozizaene (Figure Figure5151).239,240 Epi-isozizaene synthase241 catalyzes the first step of albaflavenone biosynthesis through the initial C1–C6 cyclization of FPP through an ionization-recombination-reionization sequence forming the bisabolyl cation. In contrast with the sesquiterpene cyclase reactions discussed in sections 5.2.1 and 5.2.2, which terminate upon formation of the bisabolyl cation, the reaction catalyzed by epi-isozizaene synthase proceeds through a 1,2-hydride shift, which generates the homobisabolyl cation. In turn, this cation undergoes additional carbon–carbon bond-forming reactions leading to epi-isozizaene formation as elegantly delineated through the study of isotopically labeled substrates by Cane and Lin (Figure Figure5252).242
The 1.6 Å resolution crystal structure of epi-isozizaene synthase complexed with 3 Mg2+ ions, inorganic pyrophosphate, and the benzyltriethylammonium cation reveals a single α domain enzyme characteristic of bacterial cyclases (Figure Figure5353A).243 Several features evident in this structure advance our understanding of the molecular basis of catalysis. First, 3 Mg2+ ions, 3 basic groups (R194, K247, and R338), and the phenolic hydroxyl group of Y339 comprise the molecular recognition motif for the diphosphate moiety (Figure Figure5353B) and provide the electrophilic impetus to trigger substrate ionization and catalysis. Second, the benzyltriethylammonium cation binds in the hydrophobic active site, which is encapsulated by the enzyme with a fully closed active site conformation. The quaternary ammonium cation engages F95, F96, and F198 through cation−π interactions, with the positively charged ammonium nitrogen atom approximately 5 Å away from the ring centroid of each aromatic residue (Figure Figure5353C). The structure of this complex provides a direct visualization of cation−π interactions in a terpenoid cyclase active site, demonstrating the feasibility of such interactions in stabilizing actual carbocation intermediates in catalysis. Such interactions will additionally stabilize the transition states flanking these intermediates, so cation−π interactions comprise an important strategy for transition state stabilization in a terpenoid cyclase active site.
(A) Stereoview of unliganded D99N epi-isozizaene synthase (purple) superimposed on the wild-type epi-isozizaene synthase complex with inorganic pyrophosphate (green; ligands omitted for clarity). Structural changes in helix H and the H-α-1 and ...
Comparison with the 1.90 Å resolution structure of unliganded D99N epi-isozizaene synthase reveals the molecular details of active site closure (Figure Figure5353A).243 Conformational changes in helix H, the H-α-1 loop, and the J-K loop are triggered by metal-binding and diphosphate complexation, and these conformational changes fully enclose the active site and protect it from bulk solvent. The enclosed active site contour does not unambiguously fit the three-dimensional shape of epi-isozizaene, and this may reflect the fact that the enzyme is not a high-fidelity cyclase: it generates 79% epi-isozizaene and a handful of other sesquiterpene products. Remarkably, the generation of a library of 26 single-point mutants designed to remold the active site contour triples the overall number of major and minor products generated by epi-isozizaene synthase, including the conversion of the enzyme into six different sesquiterpene synthases generating alternative major products: the F96A mutant predominantly generates (E)-β-farnesene, the F96W mutant predominantly generates zizaene, the F95H mutant predominantly generates β-curcumene, the F95M mutant predominantly generates β-acoradiene, the F198L mutant predominantly generates β-cedrene, and the F96V and W203F mutants predominantly generate (Z)-γ-bisabolene (Figure Figure5454A).244
(A) Biosynthetic manifold of wild-type and mutant epi-isozizaene synthase enzymes. The predominant cyclization product of the wild-type enzyme, epi-isozizaene, is highlighted in a cyan box, and the predominant cyclization products of mutant enzymes are ...
The ability to reprogram the cyclization cascade catalyzed by a terpenoid cyclase has profound implications for biotechnology. For example, consider F95H epi-isozizaene synthase, which generates β-curcumene as its major sesquiterpene product.244 With kcat/KM = 2600 M–1 s–1, the catalytic efficiency of this mutant enzyme is nearly 70-fold greater than that of α-bisabolene synthase (section 5.2.2).46 Since, like α-bisabolene, β-curcumene can be chemically hydrogenated to form the hydrocarbon D2 diesel fuel substitute bisabolane (Figure Figure5454B), protein engineering of a single α domain bacterial cyclase can yield a smaller and more robust catalyst than the αβγ domain plant cyclase initially utilized in this process.
Trichothecene mycotoxins such as T-2 toxin, nivalenol, and vomitoxin are occasional contaminants of grains such as wheat, oats, and corn, and fungal outbreaks in the food chain can accordingly present serious challenges to public health.245−247 For example, in 1944 moldy grain and bread contaminated with T-2 toxin caused the death of more than 10% of the population of Orenburg, Russia, leading to the exploration and weaponization of T-2 toxin and related mycotoxins for use in biological warfare.248,249 The first committed step in the biosynthesis of these mycotoxins is the cyclization of FPP to form the sesquiterpene hydrocarbon trichodiene, which is catalyzed by trichodiene synthase (Figure Figure5555).250,251
Ionization of FPP (1) and recombination of inorganic pyrophosphate yields 3(R)-nerolidyl diphosphate (2), which enables C1–C6 bond formation after reionization to yield the 4(R)-bisabolyl carbocation intermediate (4). Subsequent 1,4-hydride transfer, ...
Trichodiene synthase from the fungus Fusarium sporotrichioides is a 90-kD dimer,252 and the enzyme was cloned250 and overexpressed in E. coli.253,254 The 2.5 Å resolution crystal structures of the unliganded enzyme and its complex with 3 Mg2+ ions and inorganic pyrophosphate reveal that each monomer adopts the α domain fold of a class I terpenoid cyclase, with active sites oriented in antiparallel fashion (Figure Figure5656, panels A and B).170 The binding of 3 Mg2+ ions and inorganic pyrophosphate triggers conformational changes that fully enclose the active site; the diphosphate group is firmly bound by metal coordination interactions and hydrogen bond interactions with 3 basic residues (R182, K232, and R304) and Y305 (Figure Figure5656, panels C and D).
(A) Stereoview of trichodiene synthase. The aspartate-rich motif on helix D (magenta) and the NSE motif on helix H (red) are indicated. (B) Two monomers assemble in antiparallel fashion to form the trichodiene synthase dimer. (C) Stereoview of the trichodiene ...
Enzymological studies using isotopically labeled FPP and nerolidyl diphosphate confirm that 3(R)-nerolidyl diphosphate214 is a catalytic intermediate, as expected for the ionization-recombination-reionization sequence that enables initial C1–C6 bond formation.255,256 Single-turnover experiments indicate that FPP ionization is the rate-determining chemical step of catalysis, with product release being the overall rate-determining step of catalysis.257 Notably, incubation of trichodiene synthase with the reduced substrate analogue 7(S)-trans-6,7-dihydrofarnesyl diphosphate or its enantiomer yields product arrays that reflect formation of the corresponding 6,7-dihydronerolidyl diphosphate intermediate, thereby providing further support for 3(R)-nerolidyl diphosphate as an intermediate in the cyclization cascade.258
The tertiary 4(R)-bisabolyl cation results from C1–C6 bond formation (Figure Figure5555), and this carbocation is analogous to the 4(R)-α-terpinyl cation intermediate of monoterpene cyclization reactions (Figure Figure2929). Interestingly, just as a nonproductive binding conformation is observed for the aza analogue of the α-terpinyl cation bound in the active site of (+)-bornyl diphosphate synthase,23 so too is a nonproductive binding conformation observed for the binding of the aza analogue of the 4(R)-bisabolyl cation;178 indeed, the aza analogue of the 4(S)-bisabolyl cation binds with essentially equal affinity.259 Nonproductive binding conformations appear to be driven in part by favorable electrostatic interactions between the cationic aza moiety and the inorganic pyrophosphate anion,178,260 so here, too, the most thermodynamically stable structure corresponding to a crystalline enzyme–analogue complex at equilibrium may not necessarily correspond to the most easily accessible structure during catalysis. The trichodiene synthase reaction proceeds through a 1,4-hydride transfer, tandem 1,2-methyl migrations, and deprotonation to yield trichodiene (Figure Figure5555). Inorganic pyrophosphate is the most likely general base that mediates the final deprotonation.170,261
Interestingly, mutation of residues that coordinate to Mg2+ ions or residues that hydrogen bond with inorganic pyrophosphate results in the generation of aberrant cyclization products, more so than generated by the wild-type enzyme.262−266 Structural analysis of certain mutants reveals incomplete active site closure and an enclosed active site volume 12% larger than that of the wild-type enzyme.261 This compromises the template function of the active site and results in additional conformational flexibility for the substrate and intermediate carbocations, thereby allowing alternative carbon–carbon bond-forming trajectories that form alternative products. The proportion of alternative cyclization products increases when Mn2+ is substituted for Mg2+ in assays of mutants with substitutions in the aspartate-rich metal-binding motif,263 so metal coordination interactions must play a critical role in governing the substrate binding conformation prior to initiation of the cyclization cascade. Analysis of the trichodiene synthase mechanism using theoretical and computational chemistry provides a framework for understanding cyclization specificity and highlights the importance of the bisabolyl cation intermediate as a critical branch point in the generation of alternative sesquiterpenes.267
Selinenes contribute to the flavor and aroma of citrus fruits and German hops,268−270 and these compounds generally derive from an initial C1–C10 cyclization reaction that ultimately yields a bicyclic product (Figure Figure5757A). The C2–C3 double bond remains in the trans configuration for the cyclization reaction, so this C1–C10 cyclization reaction does not require the ionization-recombination-reionization sequence described above for C1–C6 cyclization sequences. The 2.1 Å resolution crystal structure of unliganded selina-4(15),7(11)-diene synthase from Streptomyces pristinaespiralis, also known as γ-selinene synthase or simply selinadiene synthase, reveals the characteristic single α domain fold of a bacterial terpenoid cyclase with an open active site conformation; the 2.1 Å resolution crystal structure of the complex with a substrate analogue reveals a closed active site conformation in which the diphosphate group of the analogue coordinates to 3 Mg2+ ions (Figure Figure5757B).271 Theoretical and computational chemistry studies are consistent with the proposed role of the inorganic pyrophosphate coproduct as the general base–general acid that mediates the deprotonation–reprotonation sequence shown in Figure Figure5757A, since no intramolecular proton transfers are energetically feasible in the cyclization cascade.272
(A) Cyclization mechanism of selinadiene synthase. (B) Crystal structure of selinadiene synthase complexed with a substrate analogue (yellow stick figure) showing active site helices and loops that undergo structural changes to accommodate the bound analogue ...
Dickschat and colleagues report the preparation and functional evaluation of 28 site-specific mutants of selinadiene synthase designed to probe residues that define the active site contour, residues involved in the molecular recognition of the substrate diphosphate group, and residues involved in the conformational transition required for active site closure.271 Substitution of aromatic residues indicates that these residues play important roles in stabilizing carbocation intermediates through cation−π interactions, especially F55 and F79. Conservative amino acid substitutions in the aspartate-rich metal-binding motif, D83N and D83E, do not disable the initial C1–C10 bond-forming reaction but instead compromise the second cyclization reaction mediated by the deprotonation-reprotonation sequence shown in Figure Figure5757A. Possibly, conservative mutations of metal-binding residues slightly perturb the position of metal-bound inorganic pyrophosphate, which in turn would compromise its proposed role as a general base–general acid in the cyclization sequence.
Larvae of the North American sawfly, also known as the pine sawfly, feed on the needles of infested pine trees and are responsible for occasional large-scale forest defoliation events.273 Sawfly larvae abstract the resin from pine needles and utilize it for defense against insect and animal predators such as ants, wasps, lizards, and frogs. The principal component of this defensive resin is the sesquiterpene alcohol germacradien-4-ol.274−276 Two different numbering systems for this natural product are found in the literature, so it is also known as germacradien-5-ol, as explained in the report of its total synthesis.277
Germacradien-4-ol is also generated in certain bacteria, and recent genome mining studies led to the discovery of a germacradien-4-ol synthase in Streptomyces citricolor.278 Although the wild-type enzyme did not yield crystals suitable for X-ray structure determination, the E248A mutant in its unliganded form yielded crystals that diffracted to 1.50 Å resolution.279 The side chain of E248 is a surface residue more than 14 Å away from the active site, and it was identified as a candidate for substitution to facilitate crystallization using the Surface Entropy Reduction Server.280 The structure of germacradien-4-ol synthase reveals the association of two α domains that assemble with antiparallel quaternary structure (Figure Figure5858).
Structure of the monomer (left) and dimer (right) of germacradien-4-ol synthase; aspartate-rich and NSE metal-binding motifs are red and orange, respectively. The location of the E248A mutation required to facilitate crystallization is indicated by a ...
Recent enzymological studies of the wild-type enzyme and its site-specific mutants, including the use of fluorinated and isotopically labeled substrate analogues, illuminate key features of the FPP cyclization mechanism (Figure Figure5959).279 Metal-triggered FPP ionization is followed by C1–C10 ring closure to form the germacryl cation (the C2–C3 bond of FPP remains in the trans configuration). Next, a 1,3-hydride transfer occurs with undefined stereochemistry (the deuteron in the final product is scrambled when chiral FPP deuterated at C1 is utilized as a substrate). The resulting allylic cation undergoes addition of water at the tertiary carbon to yield germacradien-4-ol; alternative minor products are generated by some mutants, as summarized in Figure Figure5959.
Definitive structure-mechanism relationships regarding the enzymatic control of the biosynthetic water molecule cannot be made in the absence of crystal structures of enzyme–ligand complexes. Although some ordered water molecules are observed in the active site of the unliganded enzyme, it is unknown as to whether any of these water molecules remain upon substrate binding to participate in catalysis. There are no obvious active site residues to serve as a general base to activate a nucleophilic water molecule, so the inorganic pyrophosphate coproduct may serve this role. Allemann and colleagues suggest that instead of reacting with a trapped water molecule, the final allylic carbocation may simply be quenched by a water molecule from bulk solvent, which may gain access to the active site through the movement of the flexible H-1α and E–F loops flanking the active site.
Generation of the bicyclic sesquiterpene aristolochene is the first committed step in the biosynthesis of several fungal toxins, including gigantenone, sporogen-AO1, bipolaroxin, and PR-toxin.281,282 Aristolochene synthase from the blue cheese mold, Penicillium roqueforti, is a 37-kD enzyme283 that has been cloned282 and overexpressed284 in E. coli, and the crystal structure of the unliganded enzyme has been determined at 2.5 Å resolution.79 Aristolochene synthase from another fungus, Aspergillus terreus, has also been cloned and overexpressed in E. coli,285 and the crystal structure of the unliganded enzyme has been determined at 2.2 Å resolution.286 The A. terreus enzyme has proven to be much more amenable than the P. roqueforti enzyme for the study of complexes with analogues of the substrate and carbocation intermediates, and a total of 10 structures of enzyme–ligand complexes have been reported at 1.86–2.40 Å resolution.24,286,287
Structural studies of A. terreus aristolochene synthase provide a useful framework for interpreting the results of early enzymological studies with stereospecifically labeled substrates,281,288,289 a mechanism-based inhibitor,290 and the modified substrate 7(R)-6,7-dihydrofarnesyl diphosphate.291 These studies are consistent with the mechanism shown in Figure Figure6060, which proceeds through initial C1–C10 bond formation to yield (S)-(−)-germacrene A (the C2–C3 bond of FPP remains in the trans configuration), which is then protonated at C6 to facilitate C2–C7 bond formation to yield the (+)-eudesmane carbocation. Stereospecific deprotonation of C8–Hβ, methyl migration, and hydride transfer yields (+)-aristolochene.
Mechanism of aristolochene synthase (PPO, diphosphate; PPO–, inorganic pyrophosphate). Some steps might proceed in concerted rather than stepwise fashion, in which case only partial positive charge would develop in place of full carbocations. ...
Additional experiments demonstrate that germacrene A is a minor product of the reaction catalyzed by P. roqueforti aristolochene synthase, consistent with the intermediacy of germacrene A in the cyclization cascade.292 Site-directed mutagenesis studies also illuminate the role of active site aromatic residues in governing reactive conformations of the substrate and carbocation intermediates, as well as stabilizing carbocation intermediates through cation−π interactions.293−298 Cation−π interactions with active site aromatic residues presumably contribute to the potency of aza analogues of carbocation intermediates as inhibitors of aristolochene synthase.299,300 Interestingly, aristolochene synthase can accommodate modified substrates with varying degrees of reactivity. For example, 2-fluorofarnesyl diphosphate, 6-fluorofarnesyl diphosphate, and 14-fluorofarnesyl diphosphate are converted to the corresponding fluorinated isomers of germacrene A,301,302 whereas 12,13-fluorofarnesyl diphosphate is an inhibitor of the enzyme.303 Such studies with modified substrates further inform our understanding of the catalytic mechanism.
Aristolochene synthase from A. terreus crystallizes as a tetramer of α domain subunits (Figure Figure6161), although native gel analysis indicates dimeric quaternary structure in solution.286 Felicetti and Cane report that enzyme activity becomes attenuated at higher protein concentrations,304 so it is possible that tetramerization is inhibitory for the cyclization reaction. The dimer is likely to be the biologically relevant species, in which the active sites are oriented in antiparallel fashion.
Analysis of the three-dimensional structure of the active site in its fully closed conformation, complexed with 3 Mg2+ ions and inorganic pyrophosphate, reveals a productlike contour; when fit with a molecule of aristolochene, both C6 and C8 atoms are perfectly oriented for stereospecific proton transfers with inorganic pyrophosphate.286 Thus, inorganic pyrophosphate is likely to be the general base–general acid that mediates the proton transfer steps, including the final deprotonation of C8–Hβ (Figure Figure6262). The subsequent crystal structure determination of the complex with an aristolochene-like aza analogue24 verifies the previously modeled286 binding conformation of aristolochene (Figure Figure6363). That the active site contour is very productlike reflects the template function of the active site. Consider that the isoprenoid substrate, FPP, is quite flexible: there are three carbon–carbon single bonds about which free rotation is possible between each 5-carbon isoprenoid unit, so the linear isoprenoid substrate is even more flexible than, for example, a polypeptide in which there are only two bond rotations per residue. Thus, the active site template must be strict as it enforces the initial binding conformation of the flexible substrate so as to ensure the correct trajectory of carbon–carbon bond formation.
Stereoview of the three-dimensional active site contours of aristolochene synthase in the unliganded conformation (top) and in the complex with 3 Mg2+ ions and inorganic pyrophosphate (bottom). Ligand binding triggers conformational changes that fully ...
All the information required to specify the final cyclization product (or products) is thus encoded in the three-dimensional contour of the active site. If this code can be deciphered then the product generated by a terpenoid cyclase can be predicted de novo. Importantly, successful attempts at predicting the product of a newly discovered sesquiterpene cyclase through computational approaches culminated in the recent discovery of cucumene synthase.305
Although a trapped water molecule comprises part of the active site contour of aristolochene synthase in the closed conformation,24 no hydroxylated products are generated by the enzyme. Moreover, mutagenesis of residues that hydrogen bond to this water molecule does not lead to the generation of hydroxylated products, so this water molecule appears to be located in a position where it simply cannot react with carbocation intermediates.25 Thus, as long as a water molecule is held in a location where it cannot react with the empty p orbital on a carbocation intermediate, then there is no risk of prematurely quenching the cyclization cascade. How terpenoid cyclases control water for organic synthesis remains a largely unanswered question.
Capsidiol is a plant phytoalexin generated in response to fungal or viral infection306−308 and derives from the oxidation of 5-epi-aristolochene, also known simply as epi-aristolochene. The crystal structure of epi-aristolochene synthase from tobacco (Nicotiana tabacum) was the first of a plant terpenoid cyclase37 and has accordingly served as a paradigm for understanding structure–function relationships in the greater family of terpenoid cyclases. The enzyme is a monomer with αβ domain architecture (Figure Figure6464), and crystal structures of complexes with FPP analogues farnesyl hydroxyphosphonate and trifluorofarnesyl diphosphate confirmed the location of the active site and illustrated how a terpenoid cyclase active site can serve as a template to chaperone the conformation of a flexible isoprenoid substrate.
Crystal structure of the epi-aristolochene synthase-farnesyl hydroxyphosphonate (FHP) complex. The substrate analogue (stick figure) binds in the active site in the α domain (orange); the β domain (blue) has no known catalytic function ...
Pre-steady-state kinetics indicate that the rate-determining step of catalysis occurs after the substrate binding step and after the ionization of FPP, indicating that a subsequent chemical step, or perhaps more likely, product release, is rate-determining.309 Structure-based mutagenesis studies310 as well as enzymological studies using a variety of fluorinated and deuterated substrates,311,312 indicate a mechanism involving metal-triggered FPP ionization, C1–C10 bond formation, and quenching of the resultantly formed germacryl cation to yield intermediate germacrene A, which after reprotonation undergoes carbon–carbon bond formation to yield the eudesmane cation intermediate. Subsequently, a 1,2-hydride transfer, 1,2-methyl migration, and final proton elimination yields epi-aristolochene (Figure Figure6565). Significantly, the chemical mechanism of epi-aristolochene formation is similar to that proposed for aristolochene synthase (section 5.2.7) but results in the formation of a different stereoisomer of aristolochene. It is remarkable that these two enzymes exert such precise conformational control over substrate and intermediates so as to generate identical carbon skeletons with alternative stereochemistries, thereby emphasizing the template function of the active site.
Even prior to the report of its three-dimensional structure, epi-aristolochene synthase served as a paradigm system for protein engineering through the construction of chimeric enzymes in which exonic domains were swapped with corresponding domains of the related sesquiterpene cyclase vetispiradiene synthase (also known as premnaspirodiene synthase) from Egyptian henbane (Hyoscyamus muticus).313 The mechanisms of these two enzymes proceed through a common eudesmane cation intermediate; a 1,2-methyl migration leads to the formation of epi-aristolochene, whereas a 1,2-alkyl migration and ring compression leads to the formation of the quaternary carbon of the vetispiradiene product (Figure Figure6565). Conformational control of this intermediate plays an important role in governing the rearrangement of the carbon skeleton.314 Interestingly, a chimeric cyclase results in an enzyme capable of generating both products. Crystal structure determinations of wild-type and chimeric cyclases, including the crystal structure of an intact enzyme–substrate complex with bound FPP, revealed structural features relevant to understanding the chemical mechanisms of both cyclases.56,315
Clearly, epi-aristolochene synthase is a versatile catalyst in terms of its potential for biosynthetic diversity, and this potential has been realized in recent studies. While epi-aristolochene synthase is a reasonably high-fidelity template for catalysis, generating 78.9% epi-aristolochene from substrate FPP, 24 additional minor side products are also generated, some of which require an initial ionization-recombination-reionization sequence to generate a cis-farnesyl cation (i.e., a “cisoid” pathway).316 Strikingly, subsequent studies show that epi-aristolochene synthase can utilize (cis,trans)-FPP as an alternative substrate with catalytic efficiency comparable to that of the natural substrate FPP (i.e., trans,trans-FPP).56,317
In a tour de force experiment that unlocked the full biosynthetic potential of epi-aristolochene synthase, 512 mutants of epi-aristolochene synthase were generated in which all possible mutations were made at nine key residues in the active site; analyzing the product arrays of the resulting 418 active enzymes, Noel and colleagues provided a definitive demonstration of how mutations influence catalysis and product specificity.318 This study not only illuminated structural and functional aspects that drive the evolution of product specificity in terpenoid cyclases but also provided critical insight relevant to the engineering of new biosynthetic pathways through synthetic biology. Undoubtedly, such potential applications will be facilitated through the computationally guided construction of epi-aristolochene synthase mutants with substantially enhanced thermostability.319
More recently, Noel and colleagues report studies of mutant enzymes that undergo self-alkylation through the introduction of a nucleophilic residue in the active site.320 Notably, alkylated enzymes are detected, many of which retain the ability to generate epi-aristolochene as the major product, but the site of alkylation is not the engineered residue. Instead, alkylation is detected at the aspartate residue in the DTE metal-binding motif and also a tyrosine in a loop flanking the active site. Interestingly, while alkylation is not detected in the wild-type enzyme when assayed with FPP at 25 °C, an alkylation product at the aspartate residue in the DTE motif is detected when the wild-type enzyme is assayed at 42 °C. X-ray crystallographic analysis of the W273E mutant at 2.43 Å resolution reveals alkylation of D444 in the DTE motif by a linear farnesyl group, so it is likely that the enzyme alkylation products observed in this study correspond to a farnesylation reaction that occurs immediately after FPP ionization. The active site mutation may slightly perturb the precatalytic binding conformation of FPP so that the intermolecular alkylation reaction successfully competes with the intramolecular cyclization reaction.
The catalytic versatility of epi-aristolochene synthase in the biosynthetic sequence of isoprenoid coupling and cyclization has also been explored in protein engineering experiments. Brodelius and colleagues demonstrate that epi-aristolochene synthase can be fused with either the N-terminus or the C-terminus of farnesyl diphosphate synthase from Artemisia annua to yield bifunctional terpenoid synthases, each of which exhibits an approximate 2-fold enhancement of epi-aristolochene generation as a function of time when incubated with GPP and IPP.321 It is unlikely that a direct channel forms between the two active sites, since the same enhancement of cyclic product generation is observed regardless of the order in which the enzymes are fused. A general proximity effect322 is more likely to account for the modest catalytic advantage observed. More recently, Chappell and colleagues show that catalytic versatility extends to the formation of terpene alkaloids, demonstrating that epi-aristolochene synthase slowly catalyzes the formation of a macrocyclic paracyclophane alkaloid from anilinogeranyl diphosphate.323
Hedycaryol is a dominant sesquiterpene alcohol found in the essential oil of Xanthocyparis vietnamensis (Vietnamese gold cypress), a recently discovered species indigenous to Northern Vietnam.324,325 Also found in flowering plants such as Camellia brevistyla, hedycaryol is thought to play a key role in the bouquet of terpenoids contributing to the floral scent so as to attract insects to the flowers for pollination.326
Hedycaryol synthase from Kitasatospora setae was recently cloned, expressed, and characterized, and the crystal structures of the unliganded enzyme and its complex with nerolidol were determined at resolutions of 2.7 and 1.5 Å, respectively.108 The enzyme adopts the single α domain fold characteristic of a bacterial sesquiterpene cyclase, with nerolidol bound deep in a hydrophobic active site where it contacts the backbone carbonyl of V179 at the helix G break (Figure Figure6666, panels A and B). As previously established in studies of squalene synthase, the exposed backbone carbonyl groups at the break in helix G may stabilize carbocation intermediates in catalysis.107
(A) Ribbon plot of hedycaryol synthase complexed with nerolidol (stick figure); helix G is green. (B) Close-up view of the helix G break in hedycaryol synthase, showing the 2.9 Å contact between the backbone carbonyl of V179 at the helix break ...
The initial C1–C10 bond-forming reaction of hedycaryol synthase is preceded by an ionization-recombination-reionization sequence through intermediate nerolidyl diphosphate, and the cyclization reaction requires formation of the cis-allylic cation prior to C1–C10 bond formation. Generation of the new C1–C10 bond results in a C11 tertiary carbocation intermediate, the helminthogermacradienyl cation,327 which is subsequently quenched by the addition of a solvent molecule to yield hedycaryol (Figure Figure6666C).
Interestingly, pH-dependent cyclization specificity is observed. At pH 8.5, FPP is converted almost exclusively into hedycaryol, with only a trace of nerolidol detected, whereas at pH 7.5, hedycaryol remains the primary product, but significant amounts of nerolidol, farnesol, β-farnesene, and α-farnesene are detected. The generation of side products reflecting incomplete ring closure suggests that a slight pH-dependent conformational change converts the active site contour into a more permissive template for the FPP cyclization reaction. A mechanism accounting for the formation of all products generated by hedycaryol synthase is found in Figure Figure6666C.
The cyclization of FPP to form (+)-δ-cadinene is the first committed step in the biosynthesis of the phytoalexin gossypol (Figure Figure6767),328−331 which is a defense metabolite in cotton that exhibits significant antimicrobial properties.332 Interestingly, gossypol has been studied as a potential cancer chemotherapeutic agent333−336 as well as a male contraceptive,337,338 so the biosynthesis of this novel C30 triterpene through the assembly and derivatization of two C15 (+)-δ-cadinene precursors has served as the focus of many studies.
Proposed mechanism of (+)-δ-cadinene synthase proceeds through an initial ionization–recombination–reionization sequence yielding a cisoid allylic cation, which then undergoes a C1–C10 cyclization reaction to yield the ...
The catalytic mechanism of (+)-δ-cadinene synthase is similar to that of hedycaryol synthase (section 5.2.9) in that the C1–C10 cyclization reaction yielding the helminthogermacradienyl cation327 is preceded by an ionization–recombination–reionization sequence through intermediate (R)-nerolidyl diphosphate.328−331 As observed for sesquiterpene cyclases such as epi-isozizaene synthase (section 5.2.3) or trichodiene synthase (section 5.2.4), (R)-nerolidyl diphosphate can also lead to initial C1–C6 cyclization reaction, and Allemann and colleagues suggest this possibility under certain conditions based on mechanistic studies using fluorinated FPP analogues with (+)-δ-cadinene synthase.339 Nevertheless, presuming that the biological substrate FPP undergoes initial C1–C10 cyclization, a 1,3-hydride shift340 yields the 3(Z),7(E)-germacryl cation, a secondary cation which rapidly undergoes intramolecular cyclization and proton elimination to yield (+)-δ-cadinene. This is a high-fidelity reaction, since the enzyme generates >98% (+)-δ-cadinene.341
The crystal structure of unliganded (+)-δ-cadinene synthase from Gossypium arboreum(342) reveals the characteristic αβ fold of a plant terpenoid cyclase, which is monomeric in the crystal and in solution.343 The crystal structure of the complex with 3 Mg2+ ions and the unreactive substrate analogue 2-fluorofarnesyl diphosphate reveals a partially open active site conformation in which the flexible substrate analogue is determined to bind with a catalytically nonproductive conformation (Figure Figure6868A). Strikingly, (+)-δ-cadinene synthase is unique among the vast majority of terpenoid cyclases in that it contains two aspartate-rich motifs DDXX(D,E) for Mg2+ complexation instead of one aspartate-rich motif and one NSE/DTE motif. Usually, the presence of two aspartate-rich motifs signals a coupling enzyme such as FPP synthase rather than a cyclase, so (+)-δ-cadinene synthase is an exception in this regard. Even so, superposition of metal-complexed structures of (+)-δ-cadinene synthase, aristolochene synthase, and FPP synthase reveals that the constellation of 3 Mg2+ ions is identical regardless of whether they are bound to a cyclase or coupling enzyme with two DDXX(D,E) motifs or DDXX(D,E)-NSE/DTE motifs (Figure Figure6868B). It is interesting to note that FPP synthase can generate cyclic sesquiterpene side products,344 so the chemical nature of metal coordination motifs in the terpenoid synthase active site do not necessarily dictate a functional difference between coupling and cyclization reactions.
(A) Stereoview showing the structure of (+)-δ-cadinene synthase complexed with 3 Mg2+ ions (gray spheres) and 2-fluorofarnesyl diphosphate (stick figure) reveals the active site in the α domain (blue) with the N-terminus of the β ...
Most recently, Allemann and colleagues demonstrate that the N-terminal polypeptide segment of (+)-δ-cadinene synthase is important for catalysis as it helps to cap the active site and protect carbocation intermediates from bulk solvent.345 Specifically, these investigators pinpoint the decapeptide segment M21RPKADFQPS, which is conserved in plant sesquiterpene cyclases as XRPXXXFXPS, as critical for this activity. Although part of this segment is disordered in the crystal structure determination, it is conceivable that this segment becomes fully ordered as the active site adopts a fully closed conformation. As evident in Figure Figure6868B, the N-terminal segment binds across the upper active site, consistent with a role in capping the active site.
The cyclization of FPP to form the tricyclic hydrocarbon pentalenene represents the first committed step in the biosynthesis of the antibiotic pentalenolactone,346−349 which inhibits glyceraldehyde-3-phosphate dehydrogenase in Gram-negative and Gram-positive bacteria.350−354 The reaction is catalyzed by pentalenene synthase, and the isolation and purification of the enzyme from Streptomyces exfoliatus UC5319 enabled the study of isotopically labeled substrates to ascertain the molecular details of the catalytic mechanism.347,355−357 Subsequent cloning and expression of the enzyme in E. coli yielded a fully functional recombinant enzyme identical in all respects to the native enzyme isolated from S. exfoliatus.358
The catalytic mechanism of pentalenene synthase as initially proposed on the basis of enzymological studies358 (path A in Figure Figure6969) involved an initial C1–C11 cyclization reaction in anti-Markovnikov fashion to form humulyl cation A, which after isomerization to humulyl cation B yields secoillud-6-en-3-yl cation D; in turn, cation D undergoes a 1,2-hydride transfer and intramolecular carbon–carbon bond formation to generate pentalenyl cation F, which yields pentalenene after proton elimination. However, insightful analysis of the pentalenene synthase reaction using theoretical and computational chemistry led Gutta and Tantillo to propose an alternative cyclization pathway from humulyl cation B to pentalenyl cation F proceeding through the 7-protoilludyl cation C.359 This alternative pathway (path B in Figure Figure6969) is consistent with all existing enzymological measurements using isotopically labeled substrates and also provided an easy rationalization for the generation of minor side product Δ6-protoilludene by certain active site mutants.360,361 The intermediacy of the 7-protoilludyl cation C was recently demonstrated by studying the isotopically sensitive branching ratio for the formation of pentalenene and Δ6-protoilludene from [6-2H]FPP as catalyzed by H309A pentalenene synthase; deprotonation of the deuterated 7-protoilludyl cation is less favorable due to the primary kinetic isotope effect and therefore yields an increased pentalenene:Δ6-protoilludene ratio.362
The crystal structure of unliganded pentalenene synthase was the first of a bacterial terpenoid cyclase and revealed the characteristic single α-domain structure topologically similar to that of FPP synthase, despite insignificant amino acid sequence identity between the two enzymes.36 The structural homology between the enzyme that generates FPP and the enzyme that cyclizes FPP indicates that these enzymes divergently evolved from a common ancestor early in the evolution of terpenoid biosynthesis, after the gene duplication and fusion event involving the core 4-helix bundle that is the precursor to the class I terpenoid synthase fold (Figure Figure2222). Although the active site of pentalenene synthase is predominantly hydrophobic, several polar residues are found at the mouth of the active site (e.g., the aspartate-rich motif and the NSE motif important for binding the trinuclear metal cluster required for catalysis) and other basic residues presumed to be important for binding the substrate diphosphate group (Figure Figure7070). Also found in this region is H309, initially proposed to serve as a general base in catalysis,36 but this role was ruled out on the basis of mutagenesis experiments.360,361 The more likely general base in the pentalenene synthase mechanism is inorganic pyrophosphate, judged feasible in quantum chemistry calculations362 and now broadly considered to serve this role.80
The possibilities for the initial cyclization of GGPP through metal-dependent ionization and departure of the diphosphate group include C1–C6, C1–C7, C1–C10, C1–C11, C1–C14, and C1–C15 bond-forming reactions, depending on which of the remaining carbon–carbon double bonds reacts with the initially formed allylic carbocation. A C3–C8 class I cyclization reaction is also possible with a bicyclic diterpene diphosphate substrate generated in a class II cyclization reaction with GGPP. As with an FPP cyclization reaction, the resulting carbocation intermediate can undergo further cyclization reactions, hydride transfers, methyl migrations, etc., before termination of the reaction by proton elimination or addition of a solvent molecule. Thus, in comparison with sesquiterpene cyclization reactions, even more diverse carbon skeletons result from diterpene cyclization reactions as a consequence of the additional carbon–carbon bond-forming trajectories available to the larger substrate.
To date, four crystal structures of monofunctional class I diterpene cyclases have been solved. Below, these cyclases are discussed in order of complexity of their initial GGPP cyclization reactions: C1–C11 (cyclooctatenol synthase), C3–C8 (ent-kaurene synthase), C1–C14 (taxadiene synthase), and a putative labdane-related diterpene cyclase (LrdC) for which the catalytic mechanism has not yet been established. The active site contour in the α domain of an enzyme with overall α or αβγ domain architecture serves as a template for catalysis by these diterpene cyclases, ensuring that substrate GGPP and subsequently formed intermediates are chaperoned through the proper sequence of reactions leading to formation of the correct product(s).
Cyclooctatin is a tricyclic diterpene with 5–8–5 ring architecture that exhibits anti-inflammatory activity through its inhibition of lysophospholipase, which is prominent in eosinophilic inflammatory reactions and catalyzes the hydrolysis of fatty acid esters to generate free fatty acids and glycerophosphate.363,364 The cyclooctatin biosynthetic gene cluster in Streptomyces melanosporofaciens MI614–43F2 contains a GGPP synthase, cyclooctatenol synthase (CotB2), and two cytochromes P450.365 The structure and stereochemistry of the 5–8–5 ring system of cyclooctatin is established in the cyclization reaction catalyzed by cyclooctatenol synthase (Figure Figure7171A).366,367
(A) Cyclooctatenol synthase (CotB2) catalyzes the cyclization of GGPP to form cyclooctat-9-en-7-ol, which undergoes subsequent oxidation by two separate cytochromes P450 (CotB3 and CotB4) to yield the lysophospholipase inhibitor cyclooctatin. (B) Active ...
The 1.64 Å resolution crystal structure of cyclooctatenol synthase was the first of a bacterial diterpene cyclase and revealed a single α-domain architecture characteristic of a class I terpenoid cyclase.368 The enzyme forms a dimer in solution and in the crystal, with active sites oriented in antiparallel fashion. The crystal structures of the unliganded wild-type enzyme and the F149L mutant reveal an open active site conformation, with a hydrophobic active site cleft flanked by aspartate-rich and NSE metal-binding motifs at the upper rim of the active site on helices D and H, respectively (Figure Figure7171A). Structural comparison with taxadiene synthase reveals an active site with decreased volume, so contrasting active site contours reflect the contrasting cyclization pathways governed by these diterpene cyclases.
Mutagenesis studies reveal that the diterpene cyclization cascade can be reprogrammed by single amino acid substitutions in the active site (Figure Figure7171B).368,369 Specifically, substitution of active site aromatic residues results in the formation of alternative products. For example, F149L cyclooctatenol synthase generates a tricyclic cyclooctatenol isomer, cyclooctat-7-en-3-ol, which reflects complete 5–8–5 ring formation with an alternative C6–C2 ring closure and quenching reaction. F107Y cyclooctatenol synthase generates the hydrocarbon cyclooctat-1,7-diene, which reflects complete 5–8–5 ring formation with an alternative proton elimination to terminate the cyclization cascade. W188H, W186L, and W288G cyclooctatenol synthases generate a bicyclic product, 3,7,18-dolabellatriene, which reflects premature derailment of the cyclization cascade by proton elimination from the first cyclic carbocation intermediate. Notably, certain dolabellatrienes exhibit antibacterial properties against multidrug resistant Staphylococcus aureus (MRSA),370 so W188H, W186L, or W288G cyclooctatenol synthases could be used for the generation of a new class of antibiotics through synthetic biology approaches.368,369
In contrast with the three mutants just described, for which the chemistry reflects a derailment of the natural cyclization pathway leading to product cyclooctat-9-en-7-ol, F107A and W186L cyclooctatenol synthases also generate cembrene A, which reflect an initial C1–C14 cyclization reaction instead of the initial C1–C11 cyclization reaction catalyzed by the wild-type enzyme and other mutants described above. The cyclization mechanisms of wild-type cyclooctatenol synthase and a wide array of active site mutants are summarized in Figure Figure7272.
Cyclization mechanisms of wild-type cyclooctatenol synthase (CotB2wt) and selected active site mutants that generate alternative cyclization products: cyclooctat-9-en-7-ol (1), 3,7,12-dolabellatriene (2), cembrane A (3), 3,7,18-dolabellatriene (4), 3,7-dolabellatrien-9-ol ...
Hundreds of kaurane diterpenoids have been identified to date, each of which adopts a core tetracyclic structure exemplified by that of ent-kaur-16-ene (Figure Figure7575).371 These compounds are generated mainly by plants and often serve as precursors to plant growth regulators. Kaurane diterpenoids also exhibit a variety of useful pharmacological properties, exhibiting antispasmodic, antiallergenic, and immunosuppressive effects; in contrast, others are embryotoxic, genotoxic, and insecticidal.372 The complex tetracyclic kauranoid skeleton is thus a naturally occurring privileged platform in biology.
The biosynthesis of the tetracyclic kaurane skeleton requires both class I and class II cyclization reactions. For example, consider the biosynthesis of ent-kaur-16-ene (Figure Figure7373). First, the class II protonation-mediated cyclization of GGPP generates ent-copalyl diphosphate (this reaction is discussed further in section 6.1). In the subsequent class I cyclization reaction, ent-copalyl diphosphate undergoes metal-dependent ionization and cyclization through several carbocation intermediates to form the tetracyclic kaurane skeleton. The class I cyclization reaction is catalyzed by ent-kaurene synthase, which in the soil bacterium Bradyrhizobium japonicum is a separate enzyme from the ent-copalyl diphosphate synthase.373 Interestingly, in other species such as fungi and the moss Physcomitrella patens, the class I and class II cyclization activities are contained in a single bifunctional enzyme.374−376
Biosynthesis of ent-kaur-16-ene, or simply ent-kaurene, in the soil bacterium Bradyrhizobium japonicum proceeds through two separate enzyme-catalyzed reactions. The first reaction is a class II protonation-initiated cyclization of GGPP to form ent-copalyl ...
The class I cyclization reaction catalyzed by ent-kaurene synthase from Bradyrhizobium japonicum is unusual in that the substrate is not the linear isoprenoid precursor GGPP but instead the partially cyclized biosynthetic intermediate ent-copalyl diphosphate. Thus, the active site of ent-kaurene synthase does not serve as a template for the binding of GGPP but instead must specifically accommodate the bicyclic diphosphate substrate to enable the formation of the remaining two rings of the tetracyclic product through an initial C3–C8 bond-forming reaction. The recently determined 1.8–2.0 Å resolution crystal structures of the unliganded enzyme, its complex with a bisphosphonate inhibitor, and the D75C mutant complexed with ent-copalyl diphosphate reveal important mechanistic inferences.377 The enzyme adopts the characteristic single α domain fold of a class I bacterial terpenoid synthase (Figure Figure7474A) and two monomers associate in roughly parallel fashion to form a dimer. While the electron density map of the D75C ent-kaurene synthase–ent-copalyl diphosphate complex is somewhat noisy (Figure Figure7474B), perhaps reflecting disorder due to the lack of bound metal ions, the structure of this open-conformation enzyme–substrate complex confirms the location of the active site and provides a view of a possible precatalytic orientation for the substrate prior to ionization and cyclization.
The taxane diterpene paclitaxel (Taxol), isolated from the bark of the Pacific yew tree Taxus brevifolia,378 exhibits potent antitumor properties by stabilizing microtubules,379,380 which at higher drug concentrations leads to cell cycle arrest and apoptosis.381,382 Recent data suggest an alternative mode of action at lower, clinically relevant concentrations in tumor cells, in which Taxol causes division of chromosomes on multipolar spindles, chromosomal missegregation, and cell death.382,383 Taxol is approved by the U.S. Food and Drug Administration for the treatment of ovarian, breast, and lung cancer and is used off-label in the treatment of several other cancers. By all measures, Taxol is highly regarded as a blockbuster drug for cancer chemotherapy.
Taxadiene synthase catalyzes the first committed step of Taxol biosynthesis and is responsible for the generation of the tricyclic hydrocarbon skeleton of Taxol with structural and stereochemical precision. Taxadiene synthase was first isolated from the pacific yew384 and cloned and expressed in E. coli as an 862-residue full-length protein.385 The amino acid sequence of the full-length protein contains a plastidial targeting sequence at the N-terminus, so the preparation of a construct lacking this sequence facilitated the overexpression of a pseudomature form of the enzyme for functional analysis.386 The major cyclization product of recombinant pseudomature taxadiene synthase is taxa-4(5),11(2)-diene (henceforth simply “taxadiene”), but minor side products taxa-4(20),11(12)-diene (~5%) and verticillene (~1%) are also generated.
The chemical mechanism of GGPP cyclization catalyzed by taxadiene synthase (Figure Figure7575) has been extensively studied using isotopically labeled substrates and fluorinated substrate analogues. Metal-triggered ionization of the substrate diphosphate group enables C1–C14 bond formation with inversion of configuration at C1, quite possibly in concert with C10–C15 bond formation, to yield the verticillyl cation intermediate.386−388 Subsequently, an unusual intramolecular proton transfer387,389,390 between C11 and the re face of C7 followed by transannular ring closure yields the taxenyl cation, which undergoes proton elimination to yield taxadiene.
The 1.82 Å resolution crystal structure of taxadiene synthase complexed with the bicyclic diterpene analogue 13-aza-13,14-dihydrocopalyl diphosphate and the 2.25 Å resolution crystal structure of taxadiene synthase complexed with the unreactive substrate analogue 2-fluorogeranylgeranyl diphosphate were the first crystal structures reported of a diterpene synthase42 and revealed αβγ domain architecture in which the class I active site resides in the α domain; the βγ domains have no known catalytic function, other than perhaps to provide the N-terminal polypeptide segment to help cap the active site upon substrate binding. This domain architecture is identical to that observed for the class II diterpene cyclase ent-copalyl diphosphate synthase (Figure Figure7676A) (section 6.1). The plastidial targeting sequence plus an additional 27 residues were deleted from the N-terminus in the taxadiene synthase construct that yielded crystals, which attenuated but did not obliterate catalytic activity. It is interesting that 13-aza-13,14-dihydrocopalyl diphosphate binds in the active site of taxadiene synthase because this analogue does not mimic any intermediates in the taxadiene synthase reaction. The binding of both diterpene analogues in the active site in the α domain is accommodated by 3 Mg2+ ions: Mg2+A and Mg2+C are coordinated by D613 and D617 in the aspartate-rich motif, and Mg2+B is chelated by N757, T761, and E765 in a hybrid NSE/DTE motif (Figure Figure7676B). Given the disorder observed for the J–K loop in these structures,42 the active site is thought to adopt a partially closed conformation. Computational techniques have recently been employed to construct a model of the fully closed conformation of taxadiene synthase.391
(A) Taxadiene synthase and ent-copalyl diphosphate synthase share a common αβγ domain architecture (α domain, blue; β domain, green; γ domain, yellow; N-terminus, magenta). The active site of taxadiene synthase ...
Although taxadiene synthase contains complete β and γ domains as found in class II terpenoid cyclases such as ent-copalyl diphosphate synthase (section 6.1), there is no corresponding catalytic function at the βγ interface of taxadiene synthase since the catalytically required general acid residue is absent. Indeed, the γ domain can be spliced out of taxadiene synthase to form a smaller αβ construct that retains a significant degree of cyclization specificity for taxadiene formation, although catalytic efficiency is attenuated.392
Given the high demand for Taxol in cancer chemotherapy, metabolic engineering approaches have been explored for the biosynthesis of taxadiene and other Taxol biosynthetic intermediates with the hope of providing a more economical and renewable source of the drug through synthetic biology.393 For example, taxadiene synthase has been utilized for taxadiene overproduction in E. coli,394 yeast (Saccharomyces cerevisiae),395Arabidopsis,396 ginseng root (Panax ginseng C.A. Meyer),397 and Artemisia annua L.398 Moreover, recent studies show that the combination of mutagenesis and pathway engineering yields an alternative approach for Taxol generation through the alternative cyclization product taxa-4(20)-11(12)-diene, which is a minor product generated by the wild-type enzyme.399 These studies suggest a promising future for the manufacture of Taxol through synthetic biology.
Recently, a labdane-related biosynthetic gene cluster was identified in the soil bacterium Streptomyces sp. K155, isolated in Valle de Chalco, Mexico, and the putative class I diterpene cyclase LrdC encoded by this cluster was crystallized.400 The crystal structure of LrdC complexed with 3 Mg2+ ions and inorganic pyrophosphate has been solved and refined at 2.36 Å resolution and the coordinates deposited in the Protein Data Bank (www.rcsb.org, PDB accession code 5A0J). The structure reveals a dimer of single α domains oriented in parallel fashion (Figure Figure7777). A full paper describing the structure and chemical mechanism of this diterpene cyclase has not yet appeared, so complete details regarding structure–function relationships are not yet available.
Crystal structure of the diterpene cyclase LrdC (PDB 5A0J).
Although sesterterpenes are comparatively rare relative to monoterpenes, sesquiterpenes, and diterpenes, examples of polycyclic C25 natural products are found in plants,401 fungi,402−408 and marine organisms such as planktonic diatoms that live in open water,409 benthic diatoms that live in surface films,409 and a variety of sponges.410−415 Many sesterterpenes exhibit useful pharmacological properties (e.g., as antitumor and anti-inflammatory agents);411,413−417 notably, an ophiobolin derivative from Bipolaris oryzae promotes the degradation of pathogenic α-synuclein in neural cells.418 The biological and pharmaceutical importance of complex sesterterpene natural products has inspired the discovery and development of novel synthetic approaches for their generation (e.g., as recently described for the total synthesis of (−)-6-epi-ophiobolin N).419
The first committed step of sesterterpene biosynthesis is the cyclization of geranylfarnesyl diphosphate (GFPP; Figure Figure11) as catalyzed by a sesterterpene cyclase. At present, no crystal structures of sesterterpene cyclases have been reported, so an analysis of structure–function relationships in this enzyme family is not currently feasible. However, these enzymes catalyze metal-dependent cyclization reactions and their amino acid sequences contain conserved metal-binding motifs characteristic of the α fold of a class I terpenoid cyclase. Furthermore, fungal sesterterpene cyclases such as ophiobolin F synthase are typically coupled with a GFPP synthase to form a bifunctional αα domain assembly.402 Similar bifunctional assemblies function in diterpene biosynthesis and are discussed in section 7.3. Sesterterpene cyclases comprise the next frontier for terpenoid cyclase structural biology and protein engineering.
Class II terpenoid cyclases initiate catalysis by protonation of a carbon–carbon π bond or epoxide moiety in an isoprenoid substrate to yield a tertiary carbocation which can then react with one of the remaining π bonds of the substrate, typically the adjacent π bond to form a 6-membered ring in concerted or stepwise fashion. The catalytic general acid is the central aspartate in the characteristic sequence motif DXDD (this motif is unrelated to the aspartate-rich DDXXD motif of class I terpenoid cyclases). The active site of a class II terpenoid cyclase is located at the interface of the β and γ domains in a protein with overall βγ or αβγ domain architecture. A βγ enzyme is a monofunctional class II cyclase, whereas an αβγ enzyme can be a monofunctional class II cyclase or a bifunctional class I-class II cyclase (bifunctional terpenoid cyclases are discussed in section 7). The remainder of this section outlines structure-mechanism relationships for class II terpenoid cyclases that utilize the C20 substrate GGPP or the C30 substrates squalene or squalene oxide.
Class II diterpene cyclases initiate catalysis by protonation of the terminal π bond of GGPP, likely in concert with C10–C15 bond formation and C6–C11 bond formation to yield a bicyclic tertiary carbocation intermediate. As with cyclization reactions catalyzed by class I terpenoid cyclases, the resulting carbocation intermediate can undergo further reactions, hydride transfers, methyl migrations, etc. before termination of the reaction by proton elimination or addition of a solvent molecule.
To date, two crystal structures of monofunctional class II diterpene cyclases have been solved. Both of them are ent-copalyl diphosphate synthases: one is a plant enzyme, the other is a bacterial enzyme. Both utilize identical chemical strategies for GGPP activation and cyclization.
The labdane-related diterpenoids comprise a family of more than 7000 natural products that serve various functions in plants, fungi, and bacteria.420,421 In plants, the labdane diterpenoid ent-copalyl diphosphate is an intermediate in the biosynthetic pathway leading to the formation of ent-kaurene and ultimately gibberellin phytohormones, which are important for plant growth and development.422−424 The class II diterpene cyclase that converts GGPP into ent-copalyl diphosphate was initially designated ent-kaurene synthetase A when first cloned from Arabidopsis thaliana,425 but this enzyme is now better known as ent-copalyl diphosphate synthase. The preparation of a pseudomature recombinant form of the enzyme lacking the N-terminal plastidial targeting sequence facilitated structural and functional analysis.426
The crystal structures of ent-copalyl diphosphate synthase complexed with substrate analogue (S)-15-aza-14,15-dihydrogeranylgeranyl thiolodiphosphate at 2.25 Å resolution and product analogue 13-aza-13,14-dihydrocopalyl diphosphate at 2.75 Å resolution were the first of a class II diterpenoid cyclase (analogue structures are shown in Figure Figure7878).44 Although the substrate analogue binds in the enzyme active site, the product analogue does not, binding instead at an interlattice site in the crystal (a product analogue with correct stereochemistry was unavailable for this study). Since the stereochemistry of the product analogue does not match that of the actual product, its exclusion from the active site reflects the precise stereochemical control that the active site imposes on the GGPP cyclization cascade to ensure that the correct product stereoisomer is formed.
Cyclization of GGPP to form ent-copalyl diphosphate is the first committed step of gibberellin biosynthesis in Arabidopsis thaliana and is catalyzed by the class II diterpene cyclase ent-copalyl diphosphate synthase. (S)-15-Aza-14,15-dihydrogeranylgeranyl ...
The overall domain architecture of ent-copalyl diphosphate synthase is αβγ, with the active site located at the βγ domain interface as indicated by the binding of the substrate analogue; although the domain architecture is identical to that of taxadiene synthase, the location of the active site is not (Figure Figure7676A). The active site cavity of ent-copalyl diphosphate synthase is quite hydrophobic in nature and its three-dimensional contour is defined by aromatic and aliphatic residues. The α domain lacks the characteristic aspartate-rich and NSE/DTE metal-binding motifs of a class I cyclase and has no known function, so it is presumed to be an evolutionary vestige.
The cyclization cascade is triggered by protonation of C14 of GGPP by general acid D379 in the D377IDD motif (Figure Figure7979).427 The superacidic anti-oriented proton of D379, approximately 104-fold more acidic than a syn-oriented proton,428 is oriented toward the substrate. Since a strong acid is required to protonate a carbon–carbon π bond (e.g., H3O+ with pKa = −1.7), the utilization of an anti-oriented carboxylic acid proton with a pKa of ~0.5 is ideal for this catalytic function. The orientation of D379 is fixed by hydrogen bond interactions with N425 and a water molecule which in turn forms a hydrogen bond with T421 (Figure Figure8080).
Cyclization of GGPP catalyzed by ent-copalyl diphosphate synthase is initiated by D379 in the DXDD general acid motif. Concerted carbon–carbon bond-forming reactions yield a bicyclic carbocation intermediate that is quenched by a solvent-mediated ...
Following formation of the bicyclic carbocation intermediate (Figure Figure7979), the cyclization cascade is terminated by proton elimination. Peters and colleagues show that the general base for this step is an active site water molecule hydrogen bonded with H263 and N322.429 Interestingly, the H263F and H263Y mutants reprogram the cyclization cascade so as to allow 1,2-hydride and 1,2-methyl shifts leading to the formation of an alternative diterpene product, kolavenyl diphosphate, so both the active site contour and the position of the general base govern product outcome.430 Cyclases that naturally generate kolavenyl diphosphate were subsequently identified to contain phenylalanine or tyrosine residues in place of the histidine residue corresponding to H263; moreover, tyrosine-to-histidine substitution results in the generation of ent-copalyl diphosphate, so this residue plays a critical role in directing the termination step of a class II terpenoid cyclase.431−433
The crystal structure of ent-copalyl diphosphate synthase complexed with substrate analogue (S)-15-aza-14,15-dihydrogeranylgeranyl thiolodiphosphate has been extended to 1.55 Å resolution, clarifying certain active site features that were ambiguous in the lower-resolution structures.434 While the tertiary ammonium cation and the thiolodiphosphate group of the substrate analogue are each disordered between two conformations in the lower resolution structure, they each adopt a single, well-ordered conformation in the higher resolution structure (Figure Figure8080). The ordering of the thiolodiphosphate group might be attributed to the presence of 1.0 mM MgCl2 in the crystallization buffer (Mg2+ is required for enzyme activity, presumably to accommodate substrate binding426); however, no well-ordered Mg2+ ions are observed in electron density maps, perhaps due to the acidic conditions of the crystallization drop (pH 5.4). Additionally, the higher-resolution structure reveals extensive hydrogen-bonded solvent networks between general acid D379 and bulk solvent through secondary channels distinct from the main active site cleft. Hydrogen-bonded solvent molecules may serve as “proton wires” to facilitate Grotthuss diffusion as the conjugate base of the general acid is reprotonated in catalysis.
The first crystal structure of a bacterial class II diterpene cyclase, recently reported at 1.80 Å resolution, reveals βγ domain architecture: the bacterial class II cyclase lacks the vestigial, nonfunctional α domain present in plant ent-copalyl diphosphate synthase (Figure Figure8181A).435 The bacterial ent-copalyl diphosphate synthase is involved in the biosynthesis of kaurenoid diterpenes platensimycin and platencin, which are thought to be promising drug leads due to their inhibitory properties against fatty acid synthase.436,437 Catalytic residues are conserved in the active sites of plant and bacterial ent-copalyl diphosphate synthase, particularly with regard to the general acid function of the DXDD motif. However, a histidine residue (H359) hydrogen bonds with general acid D313 in the bacterial enzyme, whereas an asparagine residue (N425) hydrogen bonds with general acid D379 in the plant enzyme (Figure Figure8181B). In each enzyme, the aspartic acid general acid is oriented and stabilized by hydrogen bonds for optimal catalytic function. The catalytic mechanism proposed for the bacterial enzyme is essentially identical to that proposed for the plant enzyme in Figure Figure7979.
(A) Superposition of ent-copalyl diphosphate synthase from Arabidopsis thaliana (αβγ domain architecture, yellow) and Streptomyces platensis CB00739 (βγ domain architecture, gray). (B) Superposition of ent-copalyl ...
Class II triterpene cyclases initiate catalysis by protonation of the terminal π bond of squalene or the terminal epoxide moiety of squalene oxide, in concert with a cascade of carbon–carbon bond-forming reactions ultimately yielding a multiring product. Given the size of the substrate, a 30-carbon linear achiral isoprenoid, the role of the cyclase active site as a template to enforce a productive substrate conformation is critically important. Triterpene cyclization mechanisms have been exhaustively reviewed,10 and computational approaches for studying evolutionary relationships and cyclization specificity have been described,438 so structure–mechanism relationships are only briefly reviewed here for squalene-hopene cyclase and oxidosqualene cyclase.
Hopene and related triterpene hopanoids comprise a family of pentacyclic hydrocarbons found in bacterial membranes,439,440 where they influence membrane fluidity441−443 and pH homeostasis.444 Given their exceptional chemical stability, hopanoids also serve as geological and archeological biomarkers (i.e., molecular fossils that inform the analysis of ancient rocks and sediments reflecting the evolution of life millions and even billions of years ago).445−448 Hopene is generated from the linear isoprenoid precursor squalene in a cyclization cascade catalyzed by squalene-hopene cyclase (Figure Figure8282).
Class II cyclization reaction of squalene (2) is initiated by general acid D376 (AH) and proceeds through a cascade of carbon–carbon bond-forming reactions to yield the 6–6–6–6–5 hopenyl cation (15), which undergoes ...
The 2.9 Å resolution crystal structure of squalene-hopene cyclase from Alicyclobacillus acidocaldarius was the first of a class II terpenoid cyclase,38 and the resolution of the structure was later extended to 2.0 Å as facilitated by a new crystal form.449 This monotopic membrane protein exhibits βγ domain architecture with a hydrophobic active site at the interdomain interface (Figure Figure55), accessible through a passageway that opens at the membrane. The surrounding protein surface is quite nonpolar, including helix 8 in the γ domain which anchors the protein in the membrane. Interestingly, comparison of the βγ domain architecture of squalene-hopene cyclase with the αβγ domain architecture of taxadiene synthase (Figure Figure55) reveals that the γ domain of taxadiene synthase lacks the membrane-anchoring helix. Since taxadiene synthase functions in the lumen of the plastid, it is a soluble enzyme that does not require a membrane anchor.
The 2.13 Å resolution crystal structure determination of squalene-hopene cyclase complexed with the unreactive substrate analogue 2-azasqualene illustrates the template function of the active site: 2-azasqualene binds with the conformation that would be required for the substrate to form hopene rings A–D (Figure Figure8383).450 The formation of rings A and B through C2–C7 and C6–C11 bond formation, respectively, could occur in concerted or stepwise fashion after protonation of C3 by the anti-oriented proton of general acid D376.451,452 Although C ring formation could occur in Markovnikov fashion through initial C10–C14 bond formation with subsequent ring expansion, direct anti-Markovnikov formation of the C ring through C10–C15 bond formation with stabilization of the developing secondary carbocation by W169 and/or F601, and resultant C14–C18 bond formation to yield an initial 5-membered D ring in Markovnikov fashion, would yield a 6–6–6–5 tetracyclic carbocation with rings A–C in chair conformations. Derivatives of the 6–6–6–5 carbocation intermediate are observed as minor byproducts generated by the wild-type cyclase and major products generated by mutant cyclases,453,454 strongly suggesting that this tetracyclic carbocation is an intermediate in the squalene-hopene cyclization cascade. The cyclization cascade is completed by D ring expansion to form a tetracyclic 6–6–6–6 intermediate, Markovnikov formation of the E ring to generate the pentacyclic 6–6–6–6–5 hopenyl cation, and proton elimination with the assistance of a water molecule general base to yield hopene. Alternatively, direct addition of a water molecule to the hopenyl cation yields the minor side product diplopterol (Figure Figure8282). An excellent review of the squalene-hopene cyclase mechanism has been presented by Wendt.455
Stereoview of the crystal structure of 2-azasqualene bound in the active site of squalene-hopene cyclase, showing that the ligand conformation closely matches that required for the formation of rings A–D in the squalene cyclization reaction. The ...
Given that squalene-hopene cyclase exhibits catalytic activity with C15, C20, C30, and C35 substrates (reviewed by Hauer and colleagues456,457), these enzymes have been explored as Brønsted acid catalysts for organic synthesis. For example, squalene-hopene cyclase can be engineered to serve as a general platform for the protonation and activation of alkene, epoxide, and carbonyl functional groups to enable stereoselective syntheses of various cyclohexanoids.458 The ability of squalene-hopene cyclase to perform Brønsted acid activation of various biological and nonbiological functional groups for reaction with a diverse array of nucleophiles suggests a rich future for developing new and novel catalysts for organic synthesis through protein engineering.
Also known as lanosterol synthase, oxidosqualene cyclase catalyzes the cyclization of squalene epoxide to form the steroid lanosterol (Figure Figure8484), which is the first committed step of cholesterol biosynthesis. The human enzyme shares 25% amino acid sequence identity with squalene-hopene cyclase, indicative of substantial structural homology.459 However, the characteristic general acid motif DXDD in the greater family of oxidosqualene cyclases is divergent and typically contains only the aspartic acid general acid. This residue is D455 in the human enzyme and D456 in the yeast enzyme, as confirmed by the binding of irreversible covalent inhibitors.460,461
Oxidosqualene cyclase (lanosterol synthase) initiates the cyclization of squalene oxide through protonation by general acid D455 (BH). Chair-boat-chair substrate conformation facilitates A–C ring closure reactions; following D ring closure, rearrangement ...
The 2.1 Å resolution X-ray crystal structure of human oxidosqualene cyclase complexed with product lanosterol reveals the expected βγ domain architecture with the active site located at the interdomain interface (Figure Figure8585).462 As with squalene-hopene cyclase, the active site of oxidosqualene synthase is accessed by a hydrophobic channel that leads to the membrane; the protein is anchored in the membrane by hydrophobic helix 8. Also evident in this structure is a secondary channel leading to the general acid D455. Ruf and colleagues note that solvent-mediated proton transfer through this channel may be involved in the reprotonation of D455 in catalysis.462
Human oxidosqualene cyclase is a monotopic membrane protein partially embedded in the membrane; helix 8 (green) is predominantly hydrophobic and anchors the protein in the membrane leaflet. Ordered detergent molecules and lipid fragments (blue and black ...
The binding of lanosterol in the active site of oxidosqualene cyclase (Figure Figure8686) provides a framework for understanding the complex cyclization cascade leading to lanosterol formation.462 The side chain of D455 hydrogen bonds with C456, C533, and the hydroxyl group of lanosterol, consistent with the role of D455 as the general acid catalyst. General acid-promoted epoxide ring opening and A ring formation is a concerted process.460 Steroid rings A–D interact with several aromatic residues that line the active site cavity, and these residues are proposed462 to stabilize several high-energy carbocation intermediates in the cyclization cascade (Figure Figure8484): W387, F444, and W581 may stabilize tertiary carbocation intermediates at C6 and C10, and H232 and F696 may stabilize the secondary carbocation intermediate at C14 formed by the anti-Markovnikov reaction generating ring C (Figure Figure8686).
(A) The structure of the oxidosqualene cyclase–lanosterol complex reveals that the lanosterol hydroxyl group donates a hydrogen bond to general acid D455. The side chains of aromatic residues W387, F444, and W581 may stabilize C6 and C10 carbocation ...
The D-ring closure reaction yields the protosteryl cation at C20, which undergoes proton elimination and tandem 1,2-hydride transfer and methyl migration reactions to yield lanosterol. The side chain of H232 is suitably positioned to deprotonate C9 of the protosteryl cation to terminate the reaction sequence, although the phenolic hydroxyl group of Y503 could be considered for this function as well.
Computational approaches have been used to refine the mechanistic sequence proposed in Figure Figure8484, as recently reviewed by Hess, Wu, and colleagues.463 Given that oxidosqualene cyclase is a potential target for cholesterol-lowering drugs, a deeper understanding of the molecular details of catalysis may facilitate the design of potent and selective inhibitors. Intriguingly, inhibitors of oxidosqualene cyclase may also be useful in cancer chemotherapy.464,465 Thus, continued structure–function analysis in this system may lead to new therapeutic applications.
Finally, it is interesting to note that oxidosqualene cyclase is not unique to humans; indeed, more than 80 oxidosqualene cyclases have been identified in plants, approximately one-third of which are sterol synthases and two-thirds of which are triterpene synthases that generate a variety of cyclization products.466 Thus, this enzyme family is responsible for extensive chemodiversity in the plant terpenome.
In comparison with other members of the greater family of terpenoid natural products, C35 terpenes are extremely rare: only 19 cyclic and 8 linear C35 terpenes have been identified to date, and both class I and class II terpenoid synthases are involved in their biosynthesis.467 Designated “sesquarterpenes” by Sato and colleagues,468 these compounds derive from the linear precursor heptaprenyl diphosphate, which in turn is generated from FPP and 4 IPP molecules in a processive head-to-tail coupling reaction catalyzed by the heterodimeric enzyme heptaprenyl diphosphate synthase in Bacillus subtilis (Figure Figure8787).469,470 The HepT subunit of this enzyme contains signature DDXXD Mg2+-binding motifs, so it most likely adopts the α fold of a class I terpenoid synthase. The HepS subunit is required for activity, and substrate FPP is required for HepS-HepT association.470 Heptaprenyl diphosphate is the source of the prenyl side chain of menaquinone-7, which functions in electron transport. In plants, polyprenyl diphosphate synthases have been identified that generate a range of isoprenoid diphosphates, including heptaprenyl diphosphate.471,472
Sesquarterpene biosynthesis in B. subtilis begins with the condensation of FPP and 4 IPP molecules in a chain elongation reaction catalyzed by the heterodimeric enzyme heptaprenyl diphosphate synthase (HepS-HepT). YtpB catalyzes the cyclization of heptaprenyl diphosphate ...
Elegant studies by Sato and colleagues show that Bacillus and Mycobacterium species contain sesquarterpene synthases, as recently reviewed.467 Two C35 biosynthetic enzymes have been identified in each species, and the enzymes of B. subtilis are particularly well-characterized. One is the terpenoid cyclase tetraprenyl-β-curcumene synthase (YtpB), which catalyzes the ionization and cyclization of heptaprenyl diphosphate to form a C35 product containing a single β-curcumene ring (Figure Figure8787).468 This reaction resembles that of a class I terpenoid synthase. Although a naturally occurring β-curcumene synthase has not yet been isolated, F95H epi-isozizaene synthase, a class I terpenoid cyclase, generates β-curcumene as its predominant product.244 YtpB does not contain the signature DDXXD or NSE metal-binding motifs found in epi-isozizaene synthase; however, it contains D88NLCD and D249YLIDQEED segments reminiscent of metal-binding motifs in UbiA (section 3.2). Thus, YtpB might be a class I terpenoid synthase. Clearly, an X-ray crystal structure determination is required to clarify this possibility.
The second sesquarterpene cyclase identified in B. subtilis is tetraprenyl-β-curcumene cyclase (SqhC), which generates tetracyclic sporulene 4 shown in Figure Figure8787.468,473 This C35 cyclase SqhC is highly similar to authentic C30 squalene cyclases and accordingly contains the signature general acid motif DXDD of a class II terpenoid synthase.473 Although no crystal structure is available for this enzyme at present, SqhC presumably adopts the βγ domain architecture as observed for the squalene-hopene cyclase (Figure Figure55).38 Assays with cell-free extracts containing recombinant SqhC conclusively demonstrate that this enzyme is a class II terpenoid cyclase, so the protonation of the terminal C=C bond of the isoprenoid substrate in Figure Figure8888 is achieved by the central aspartic acid general acid in the DXDD motif.468 Interestingly, SqhC from Bacillus megaterium (but not SqhC from B. subtilis) can also utilize C30 squalene as a substrate (Figure Figure8888).474 Further sesquarterpene structural diversity emanates from nonenzymatic oxidative and thermal degradation reactions involving SqhC cyclization products (Figure Figure8787).475
Tetraprenyl-β-curcumene synthase (SqhC) from Bacillus megaterium generates a tetracyclic C35 sporulene from tetraprenyl-β-curcumene in a class II terpenoid synthase reaction (top). This enzyme also catalyzes the cyclization of C30 squalene ...
Terpenoid biosynthesis provides occasional examples of reaction sequences in which tandem diterpene cyclization or coupling-cyclization reactions are catalyzed in two distinct active sites on the same protein. Even in the absence of a direct channel between active sites, such systems could potentially provide a catalytic advantage through a proximity effect (“proximity channeling”)322 but only if their active sites are appropriately oriented with regard to each other. In the case of an oligomeric protein, a catalytic advantage could result from cluster channeling: although the probability might be low that a product of one reaction will be the substrate of the second reaction in a bifunctional enzyme if the active sites are not appropriately oriented, the probability that the second reaction will be catalyzed by one of the multiple subunits in an oligomeric assembly or aggregate can be higher (Figure Figure8989).476
Hypothetical enzymes E1 and E2 catalyze successive biosynthetic reactions, such that E1 converts the substrate into an intermediate, which in turn serves as a substrate for E2 to generate the final product. The active sites of E1 and E2 can be connected ...
Here, three examples of bifunctional diterpene synthases are reviewed. The first, geosmin synthase, is a class I–class I sesquiterpene cyclase in which channeling does not occur. The second, abietadiene synthase, is a class I–class II diterpene cyclase with αβγ domain architecture in which channeling is similarly not observed. The third, fusicoccadiene synthase, is a hexameric class I–class I diterpene synthase with (αα)6 domain architecture in which cluster channeling is implicated.
The characteristic odor of freshly turned earth derives from the volatile sesquiterpene geosmin,207,477 which is generated by various soil-dwelling Streptomyces species.478,479 With an extremely low detection threshold of less than 10 parts per trillion, geosmin is a commonly occurring contaminant of water, wine, and fish.480−483 While geosmin is not believed to be toxic to humans, the musty taste that it imparts to contaminated food and water makes its detection and elimination an environmental priority.
Geosmin is a noncanonical sesquiterpene in that although it derives from the universal 15-carbon precursor FPP, it only contains 12 carbons. Thus, the FPP cyclization reaction must also include a fragmentation reaction. Geosmin synthase from S. coelicolor is a 726-residue protein containing two class I terpenoid synthase α domains,484,485 both of which are required for geosmin biosynthesis.486,487 Thus, geosmin synthase is a bifunctional terpenoid synthase that adopts αα domain architecture. The N-terminal domain catalyzes the Mg2+-dependent cyclization of FPP initially proceeding through a C1–C10 bond-forming reaction to yield two cyclic sesquiterpenes, germacradienol (85%) and germacrene D (15%).488,489 The C2–C3 double bond of FPP remains in the trans configuration for the initial C1–C10 bond-forming reaction. After fully dissociating from the N-terminal domain active site into solution, product germacradienol rebinds in the C-terminal domain active site, where it undergoes a protonation-dependent cyclization-fragmentation reaction yielding C12 geosmin and C3 acetone (Figure Figure9090).490,491
Cyclization of FPP to form germacrene D and germacradienol (black arrow) is catalyzed by the N-terminal α domain of geosmin synthase. The subsequent cyclization and retro-Prins fragmentation of germacradienol (green arrows) are catalyzed by the ...
Full-length geosmin synthase proved to be refractory to crystallization, but a construct of the unliganded N-terminal domain containing residues 1–366 yielded crystals that diffracted X-rays to 3.2 Å resolution; another N-terminal domain construct consisting of residues 1–338 complexed with 3 Mg2+ ions and the bisphosphonate inhibitor alendronate yielded crystals that diffracted to 2.11 Å resolution.492 Comparison of the unliganded and liganded structures reveals the structural changes and disorder–order transitions that accompany ligand binding and active site closure for catalysis (Figure Figure9191A). The trinuclear metal cluster is coordinated by the first residue in the aspartate-rich motif, D86, the NSE motif on helix H, and the bisphosphonate moiety of alendronate. Three basic residues (R184, R236, and R325) additionally form hydrogen bonds with the bisphosphonate group (Figure Figure9191B). Both metal coordination and hydrogen bond interactions activate the FPP diphosphate group for catalysis.
(A) Structural changes triggered by the binding of the bisphosphonate inhibitor alendronate (stick figure) and 3 Mg2+ ions (magenta) to the N-terminal domain of geosmin synthase. The aspartate-rich and NSE metal-binding motifs are red and orange, respectively. ...
The fragmentation step in geosmin biosynthesis is unprecedented in terpenoid cyclase reactions. Intriguingly, even though the cyclization–fragmentation reaction catalyzed by the C-terminal domain does not involve FPP or inorganic pyrophosphate, this reaction is Mg2+-dependent.487 The chemical and/or structural function of the metal ion in this domain remains unknown. Curiously, however, the amino acid sequence of the C-terminal domain reveals conservation of metal-binding motifs as well as basic residues that could interact with inorganic pyrophosphate, even though it is not required for the fragmentation reaction.487 A homology model of the C-terminal domain provides useful structural insight on domain assembly in the geosmin synthase monomer, as fit to a low-resolution molecular envelope determined by small-angle X-ray scattering.492 However, this model does not clarify the puzzling features regarding the metal requirement for catalysis in the C-terminal domain. The answer to this question must await the crystal structure determination of a full-length αα domain enzyme.
(−)-Abietic acid is one of the main components of the oleoresin secreted by the grand fir (Abies grandis) in response to the bite of a bark beetle. Oleoresin is a complex mixture comprised mainly of volatile monoterpenes (turpentine) and diterpene resin acids;493 bactericidal and fungicidal terpenoids are also generated.494 The monoterpenes ultimately evaporate, leaving the diterpene resin acids behind to harden into a solid rosin, thereby sealing the wound. The first committed step in the biosynthesis of (−)-abietic acid is catalyzed by abietadiene synthase,495,496 which catalyzes the cyclization of GGPP to form (−)-abieta-7(8),13(14)-diene, or simply abietadiene. The overall cyclization sequence proceeds through two consecutive reactions catalyzed by class II and class I active sites in abietadiene synthase.
The mechanistic details of both reactions have been determined through enzymological studies of a pseudomature form of the enzyme lacking the N-terminal plastidial targeting sequence.192 The first reaction catalyzed is the class II cyclization reaction of GGPP, protonation of the terminal π bond of GGPP by general acid D404 initiates a cyclization cascade forming (+)-copalyl diphosphate (a different stereoisomer from ent-copalyl diphosphate described in sections 6.1.1 and 6.1.2).497−499 This biosynthetic intermediate diffuses into bulk solvent and then rebinds in the class I active site of the enzyme; since free (+)-copalyl diphosphate is detected in solution in kinetic assays, there is no evidence for channeling in this bifunctional cyclase.497 In the class I active site, metal-triggered500 ionization of (+)-copalyl diphosphate and cyclization yields a tricyclic carbocation501 that subsequently undergoes intramolecular proton transfer, 1,2-methyl migration, and deprotonation to yield abietadiene or regiomeric elimination products levopimaradiene and neoabietadiene (Figure Figure9292).192,501,502 Although the class I active site in the α domain and the class II active site at the βγ domain interface are independent, truncation studies demonstrate that these active sites cannot be separated into functional stand-alone α and βγ domain assemblies, indicating significant interdomain contact and structural stabilization in the intact αβγ assembly of the full-length enzyme.43
Cyclization of GGPP (1) is initiated in the class II active site of abietadiene synthase by general acid D404 (B:H) to yield bicyclic carbocation 1a, which undergoes proton elimination to yield (+)-copalyl diphosphate (2). This intermediate then diffuses ...
The 2.30 Å resolution X-ray crystal structure of unliganded abietadiene synthase was the first of a bifunctional terpenoid cyclase and confirmed αβγ domain architecture with fully functional class I and class II active sites (Figure Figure9393).45 The active site in the α domain contains the aspartate-rich DDXXD and NTE metal-binding motifs on helices D and H, respectively, and the active site at the βγ domain interface contains the general acid motif DXDD. Interestingly, molecular dynamics simulations suggest that conformational changes of loop 482–492 in the class II active site play an important role in governing the conformation of substrate GGPP. Structural comparison of abietadiene synthase with ent-copalyl diphosphate synthase reveals “out” and “in” conformations for this loop, respectively (Figure Figure9494).45 Since the “loop-in” conformation observed in ent-copalyl diphosphate synthase accommodates a nonproductive binding conformation for a bound substrate analogue, the “loop-out” conformation may be required to facilitate the formation of a catalytically productive enzyme–substrate complex.
Loop 482–492 (green) adopts an “out” conformation in abietadiene synthase, whereas the corresponding loop of ent-copalyl diphosphate synthase adopts an “in” conformation. The “out” conformation of ...
The crystal structure of abietadiene synthase provides a valuable framework for understanding protein engineering experiments that reprogram the cyclization cascade to generate alternative diterpene products. For example, the H348A substitution in the class II active site allows the final carbocation intermediate to be quenched by solvent to yield 8α-hydroxy-copalyl diphosphate;503 in the wild-type enzyme, H348 serves as the catalytic base in conjunction with a hydrogen-bonded tyrosine residue.504 In the class I active site, the A723S substitution converts abietadiene synthase into a pimaradiene synthase.505 Certain residues in each cyclase active site appear to be “hot spots” for reprogramming the respective cyclization cascades. Chemical dynamics simulations of the class I reaction suggest that the enzyme active site structure plays a vital role in steering the reaction through one particular pathway along a bifurcated potential energy surface, further demonstrating the critical role of the active site contour as a chaperone for cyclization.506
Fusicoccin A (Figure Figure9595) is a diterpene glucoside507−509 generated by the fungus Phomopsis amygdali, which causes the plant disease constriction canker as first discovered in peach orchards nearly a century ago.510−512 Fusicoccin A is also an emerging new lead for cancer chemotherapy due to its antitumor activity,513 which derives from its ability to modulate protein–protein interactions involving the 14–3–3 protein514−516 as well as its ability to inhibit focal adhesion kinase.517 The first committed step of fusicoccin A biosynthesis is the cyclization of GGPP to form fusicoccadiene, a tricyclic diterpene hydrocarbon with an unusual 5–8–5 ring system. This reaction is catalyzed by fusicoccadiene synthase, which exhibits αα domain architecture based on analysis of its amino acid sequence: the C-terminal α domain is a coupling enzyme that catalyzes the chain elongation reaction of DMAPP and three IPP molecules to generate GGPP, and the N-terminal domain is a cyclase that converts GGPP into fusicoccadiene (Figure Figure9595).518 The overall amino acid sequence identity between the N-terminal and C-terminal α domains is 19%.
C-terminal α domain of fusicoccadiene synthase catalyzes the coupling of DMAPP with three molecules of IPP (blue box), and the resulting molecule of GGPP is cyclized in the N-terminal α domain to yield fusicoccadiene (red box). Additional ...
Although full-length fusicoccadiene synthase was refractory to crystallization, the preparation of individual N-terminal and C-terminal α domain constructs enabled their crystallization and structure determination at 2.30 and 2.43 Å resolution, respectively.95 The crystal structures of the C-terminal GGPP synthase domain in the unliganded and liganded states reveal substrate-induced conformational changes that enclose the active site, facilitated by the binding of 3 Co2+ ions and the bisphosphonate inhibitor pamidronate (Figure Figure9696A). As is characteristic for a terpenoid synthase that catalyzes a coupling reaction, metal ions are coordinated by two aspartate-rich motifs on helices D and H, D474DFQD, and D605DYQN, respectively. The bisphosphonate moiety of the inhibitor is additionally stabilized by hydrogen bonds with 3 basic residues, R483, K619, and K629; an additional hydrogen bond is observed in some monomers with K561. The GGPP synthase domain adopts hexameric quaternary structure similar to that first observed for human GGPP synthase.88
(A) Crystal structure of the C-terminal GGPP synthase domain of fusicoccadiene synthase. Aspartate-rich DDXXD metal-binding motifs on helices D and H (red) coordinate to 3 Co2+ ions (magenta spheres) along with the bisphosphonate inhibitor pamidronate ...
The crystal structure of the N-terminal cyclase domain complexed with 3 Mg2+ ions and pamidronate reveals a monomeric enzyme in which the metal ions are coordinated by an aspartate-rich motif on helix D, D92DVTD, and the NSE motif on helix H, N232DIWSWPKE, as is characteristic for a terpenoid synthase that catalyzes a cyclization reaction (Figure Figure9696B). The bisphosphonate moiety of the inhibitor is additionally stabilized by hydrogen bonds with 3 basic residues, R188, K239, and R325, as well as a tyrosine residue, Y326. Locked in the closed conformation by metal coordination and hydrogen bond interactions with the inhibitor bisphosphonate moiety, the active site of the N-terminal cyclase domain is fully enclosed and sequestered from bulk solvent. The three-dimensional contour of the active site is quite complementary to the three-dimensional structure of product fusicoccadiene, which enables the modeling of the enzyme–product complex (Figure Figure9696C). This model illustrates the critical role of the active site contour as a template for catalysis. The template also enforces an unusual transannular proton transfer in the cyclization cascade,519 which generates fusicoccadiene as a major product and (+)-δ-araneosene as a minor product (Figure Figure9797).
Given that the C-terminal GGPP synthase domain of fusicoccadiene synthase crystallizes as a hexamer, and given that full-length fusicoccadiene synthase is a hexamer in solution as determined by analytical ultracentrifugation, the crystal structure of the GGPP synthase hexamer was used as a starting point to guide the interpretation of a low-resolution molecular envelope of full-length fusicoccadiene synthase determined by small-angle X-ray scattering.95 An optimal fit to the molecular envelope required a slight, ~5 Å dissociation of the hexamer into 3 dimers, which then facilitated the placement of 3 crystallographic dimers of the cyclization domain, all while maintaining the original D3 symmetry of the GGPP synthase hexamer (Figure Figure9898). In the model, the C-terminus of the cyclization domain and the N-terminus of the GGPP synthase domain are separated by ~25 Å, which would be connected by the 60-residue interdomain linker that was excluded from both constructs. The model of hexameric fusicoccadiene synthase shown in Figure Figure9898 provides a plausible starting point for studying the catalytic importance of quaternary structure in this system.
Model of the fusicoccadiene synthase hexamer with D3 symmetry fit into the low-resolution ab initio molecular envelope calculated from small-angle X-ray scattering data. Dimers of the GGPP synthase domains and cyclase domains are green and blue, respectively. ...
In particular, the close assembly of 6 GGPP synthase active sites and 6 GGPP cyclase active sites raises the possibility of cluster channeling in the hexamer: the GGPP product of one GGPP synthase domain may diffuse within the hexamer to any one of the 6 GGPP cyclase domains instead of diffusing out to bulk solvent. A competition experiment with fusicoccadiene synthase and taxadiene synthase demonstrates that when equimolar concentrations of both diterpene cyclases are incubated directly with GGPP, the fusicoccadiene/taxadiene product ratio is 4.3:1, reflecting that the catalytic efficiency of fusicoccadiene synthase is greater than that of taxadiene synthase. However, when the mixture is incubated with DMAPP and IPP, such that the only source of GGPP is that generated by the GGPP synthase domain of fusicoccadiene synthase, the fusicoccadiene/taxadiene product ratio jumps to 46:1. Hardly any taxadiene is generated when both cyclases compete for the GGPP generated by fusicoccadiene synthase, suggesting that most of the GGPP generated remains associated with the hexamer and utilized for cyclization to fusicoccadiene.392
While carbocations reign supreme in the terpenoid synthase and cyclization reactions discussed up until now, it is fitting to conclude this Review with two unusual examples of anionic isoprenoid cyclization reactions triggered by substrate reduction or oxidation. Notably, these reactions do not involve linear isoprenoid diphosphate precursors but instead involve GPP derivatives primed for redox chemistry.
Cannabis has been a component of the pharmacopoeia for several millennia and is mentioned in the ancient herbal medicine text, Pên-ts’ao Ching, attributed to Emperor Shen Nung, who is regarded as the legendary father of agriculture and herbal medicine (ca. 2700 B.C.E.):520 “...if taken in excess will produce hallucinations. If taken over a long-term, it makes one communicate with spirits and lightens one’s body.” More recently, cannabinoids are being used, or are being considered for use, in the treatment of glaucoma, epilepsy, multiple sclerosis, and other debilitating diseases.521 The cannabinoid Δ1-tetrahydrocannabinol (THC; also known as Δ9-tetrahydrocannabinol using a different numbering convention) is the main psychoactive ingredient found in the Cannabis sativa plant.522 It is not often appreciated that this plant contains more than 120 different cannabinoids that exhibit diverse and potentially beneficial physiological effects distinct from the psychoactive effects of THC.523,524
The first step of cannabinoid biosynthesis is the prenylation of olivetoic acid by a Mg2+-dependent prenyltransferase, geranylpyrophosphate:olivetolate geranyltransferase, to yield cannabigerolic acid (Figure Figure9999).525 Since this enzyme is soluble and is not believed to be membrane-associated, it may be related to the ABBA prenyltransferases (section 3.1),119,120 although this has not been established. Cannabigerolic acid then undergoes a complex cyclization cascade to yield Δ1-tetrahydrocannabinolic acid (THCA) in a reaction catalyzed by THCA synthase.526,527 Nonenzymatic decarboxylation of THCA yields the psychoactive cannabinoid THC.
Biosynthesis of Δ1-tetrahydrocannabinolic acid (THCA).
The recently determined crystal structure of THCA synthase illuminates key structure–mechanism relationships for the oxidative monoterpene cyclization reaction of cannabigerolic acid.528 The FAD cofactor is covalently bound to H114 and C176 at the interface of two globular domains designated I and II, where the nucleotide moiety of FAD binds between subdomains Ia and Ib (Figure 100). Phylogenetic and structural studies indicate that THCA synthase is most closely related to another FAD-dependent enzyme, berberine bridge enzyme (39% sequence identity), which functions in benzophenanthridine biosynthesis in the California poppy, Eschscholzia californica.529
Although the crystal structure of THCA synthase was determined in the absence of bound substrate, substrate analogue, or product, analysis of the native enzyme complexed with cofactor FAD suggests that active site residue Y484 serves as a general base in catalysis, and mutagenesis data confirm that this residue is absolutely required for catalysis.528 This discovery is the basis of a tentative proposal for the cyclization mechanism, in which proton abstraction by the phenolate anion of Y484 triggers substrate oxidation and intramolecular oxo-Diels–Alder cyclization to form the tricyclic product THCA (Figure 101). However, this mechanistic proposal remains to be fully developed, since it does not account for the trans–cis isomerization of the isoprenoid double bond in product THCA.
Proposed catalytic mechanism of THCA synthase based on crystal structure analysis. Y417 and H292 are proposed to function in substrate binding, and Y484 is the proposed general base. Cofactor flavin adenine dinucleotide (FAD), substrate cannabigerolic ...
Iridoids are bicyclic monoterpenes that function in communication and defense in plants and insects.530,531 The core skeleton of an iridoid typically consists of a 5-membered cyclopentane ring fused with a 6-membered pyran ring, which is further derivatized and often glycosylated to yield a family of approximately 600 distinct natural products. Many iridoids exhibit useful pharmacological properties, so these compounds figure prominently in medicinal chemistry.532
The cyclization substrate for iridoid biosynthesis is not the universal monoterpene precursor GPP but instead is 8-oxogeranial, which derives from GPP through hydrolysis to geraniol and subsequent oxidation at C1 and C8 to yield the dialdehyde species (note that 8-oxogeranial is sometimes referred to as 10-oxogeranial based on a different atomic numbering convention).533,534 The reductive cyclization of 8-oxogeranial with NADPH generates nepatalactol, which is in equilibrium with iridodial (Figure 102A).535−538
(A) Iridoid synthase from Catharanthus roseus (CrISY) catalyzes the NADPH-dependent reductive cyclization of 8-oxogeranial to form iridodial, which is in equilibrium with nepatalactol. (B) The plant enzyme progesterone 5β-reductase (P5βR) ...
The recently determined crystal structures538,539 of iridoid cyclase from Catharanthus roseus reveal close structural homology with the crystal structure of progesterone-5β-reductase, an NADPH-dependent reductase that catalyzes the reduction of the double bond in the α,β-unsaturated ketone of a steroid substrate (Figure 102B).540 The structure and function of these two enzyme families are closely linked in plants.541 O’Connor and colleagues elegantly demonstrate that the cyclization reaction catalyzed by iridoid synthase proceeds through an intramolecular Michael cyclization reaction; Rauhut-Currier and imine-assisted cyclization mechanisms are ruled out on the basis of the active site structure and mutagenesis studies, and a hetero Diels–Alder reaction is ruled out on the basis of reactivity differences of carefully designed substrate analogues (Figure 102C).536−538
The binding of substrate and analogues in the active site of iridoid synthase along with cofactor NADP+ provides important mechanistic inferences.538,539 The crystal structure of the ternary complex with triethylene glycol carboxylic acid reveals an oxyanion hole comprised of Y178 and the backbone NH groups of I145 and K146 (Figure 103A). The structure of the ternary complex with geranic acid, which mimics the enolate intermediate resulting from hydride transfer, shows that the reactive aldehyde carbonyl of the substrate can bind in the oxyanion hole while holding the β-carbon of the substrate over the pro-S hydride of the nicotinamide ring of NADPH (Figure 103B). Interestingly, geranic acid adopts an extended conformation that is nonproductive with respect to the Michael cyclization reaction following the initial hydride transfer step in catalysis.538 This is consistent with the structure of the enzyme–substrate complex.539 The observation of “open” and “closed” active site conformations involving the G150-D162 loop (Figure 104) suggests that transition of this loop to an “open” conformation would facilitate Michael cyclization to iridodial, as confirmed by molecular modeling (Figure 103C).
(A) The binding of triethylene glycol carboxylate (TEG) in the active site of the iridoid synthase-NADP+ complex reveals that a carboxylate oxygen receives hydrogen bonds from the backbone NH groups of I145 and K146 and the hydroxyl group of Y178. These ...
In the iridoid synthase-NADP+-GEA complex (gray), the G150-D162 loop partially encloses the active site. In order to accommodate iridodial binding, F152 and its associated loop must undergo a conformational change to an open active site conformation, ...
Given the vast chemodiversity of the terpenome, it is remarkable that only three modular protein domains, α, β, and γ, with γ = β′, are responsible for generating the (poly)cyclic carbon skeletons of more than 80000 terpenoid natural products. There is no sequence identity or structural homology between the α domain fold and the β domain fold, so these folds presumably evolved independently. Evolutionarily divergent protein folds observed in the greater family of terpenoid synthases have been systematically classified.40
The α domain fold derives from gene duplication and fusion of a primordial 4-helix bundle protein;41 characteristic metal-binding motifs are conserved at equivalent positions on topologically related helices D and H, suggesting that the original 4-helix bundle ancestor was a metal-binding protein. The crystal structure of avian FPP synthase provided the first view of the α fold,58 which was unexpectedly observed in the first crystal structures of class I terpenoid cyclases, bacterial pentalenene synthase, and plant epi-aristolochene synthase.36,37 The α domain is also adapted for function as an integral membrane protein, as recently observed in the crystal structures of UbiA prenyltransferases that function in ubiquinone biosynthesis.41,128 The α domain fold is thus utilized for terpenoid coupling and/or cyclization reactions in all forms of life, where it is found in α, αα, αβ, and αβγ domain assemblies.
Bifunctional terpenoid synthases such as fusicoccadiene synthase contain linked cyclase and coupling domains with αα domain architecture. The amino acid sequences of metal-binding motifs generally distinguish a terpenoid cyclase from a prenyltransferase, although there can be slight variations in metal-binding motifs in all terpenoid synthases.
The β domain is found in terpenoid cyclases as an αβ assembly as first observed in epi-aristolochene synthase,37 as a βγ assembly as first observed in squalene-hopene cyclase,38 or as an αβγ domain assembly, as first observed in taxadiene synthase,42ent-copalyl diphosphate synthase,44 and abietadiene synthase.45 In class I αβ domain assemblies, the β domain has no direct catalytic function other than to provide an N-terminal polypeptide segment to help cap the closed conformation of the active site in the α domain. This may be an evolutionary vestige from a bifunctional copalyl diphosphate synthase-kaurene synthase that was the ancestor to all plant terpene synthases,17 given that the γ domain is readily lost.59 In class II βγ or αβγ domain assemblies, the active site is at the βγ domain interface and the catalytic general acid is contained in the β domain.38,44 The γ domain diverged from the β domain8 and is the location of a membrane-anchoring component in the monotopic membrane proteins squalene-hopene cyclase and oxidosqualene cyclase.38,462 Interestingly, the β domain fold is conserved in other enzymes that utilize terpenoid substrates, such as protein farnesyl transferase and linalool dehydratase isomerase.542,543 Nature derives impressive functionality from these basic protein folds that are readily adapted for function in the cell cytosol or the membrane. In a sense, the α, β, and γ domains comprise privileged platforms for biosynthesis.
With regard to the chemistry of catalysis, class I and class II terpenoid synthases utilize different chemical strategies to initiate carbocation formation in isoprenoid substrates: a class I cyclase or prenyltransferase utilizes 3 metal ions and 3 basic residues to trigger the departure of the substrate diphosphate group to generate an allylic cation, whereas a class II cyclase utilizes an aspartic acid general acid to protonate an isoprenoid π bond and generate a tertiary carbocation. Strikingly, both chemical strategies are in play in the bifunctional diterpene cyclase abietadiene synthase.45
Despite differences in substrate activation, however, both class I and class II terpenoid cyclases serve critical functions as templates for catalysis. In each cyclase active site, the flexible polyisoprenoid substrate must be fixed in place prior to initial carbocation formation to ensure that the proper trajectory of carbon–carbon bond-forming reactions ensues. The correct sequence of carbon–carbon bond forming reactions in the cyclization cascade is directed by the precatalytic binding conformation of the substrate and the accessible conformation(s) of subsequently formed carbocation intermediates, so the fidelity of the cyclization pathway is encoded in the three-dimensional contour of the cyclase active site. The active site contour can be remolded through site-directed mutagenesis to introduce new reaction pathways that generate alternative products. Accordingly, there is great promise for the use of terpenoid synthases in synthetic biology for the manufacture of commercially important chemical compounds, such as biofuels and drugs.167−169,394,399 The ability to do so predictably (i.e., to predict the structure of a terpenoid cyclization product based solely on analysis of the structure of the terpenoid cyclase active site) is a grand challenge for computational protein engineering, and a successful first step in this regard has been recently reported.305
The active sites of both class I and class II terpenoid cyclases contain at least 2–3 aromatic residues capable of stabilizing high-energy carbocation intermediates and their flanking transition states through cation−π interactions. Additional stabilization of carbocation intermediates in class I cyclases may derive from the helix dipole at the break in helix G, as first suggested for the coupling enzyme squalene synthase107 and subsequently noted for hedycaryol synthase108 and selinadiene synthase.271 Additionally, the negative electrostatic potential of the inorganic pyrophosphate coproduct provides electrostatic stabilization of carbocation intermediates. Inorganic pyrophosphate may also serve widely as a general base to terminate the reaction sequence in class I terpenoid cyclization and coupling reactions,3,75,80 and a water molecule may serve this function in class II terpenoid cyclases.429 For class I or class II cyclases in which a water molecule adds to the final carbocation intermediate to terminate the reaction, the water molecule must be highly controlled so that it does not prematurely quench the sequence of carbocation intermediates in the cyclization cascade.
In closing, it is interesting to consider the glimpse of assembly line biosynthesis in more complex terpenoid synthase structures containing multidomain bifunctional assemblies such as abietadiene synthase (αβγ domain architecture),45 or oligomeric multidomain bifunctional assemblies such as fusicoccadiene synthase [(αα)6 domain architecture and oligomeric assembly].95 Reaction flux in such systems can be enhanced through cluster channeling, as recently demonstrated for fusicoccadiene synthase.392 Moreover, combinations of class I and class II cyclases lead to the generation of cyclic diterpenes that are “new to nature”,544,545 and computational analyses of monoterpene carbocation structures indicate that known cyclic terpenoids represent just a small subset of the theoretically maximum number of possible carbon skeletons that can be formed from the linear achiral precursor GPP.546 Thus, there is tremendous untapped potential in terpenoid chemodiversity that can be explored through protein engineering. Future studies will undoubtedly continue to illuminate the versatility and utility of terpenoid synthase structure and function, which promises to play an increasingly prominent role in synthetic biology approaches for the manufacture of carbon-based materials.
I am grateful to Drs. David Cane, Golda Harris, Sarah O’Connor, Travis Pemberton, Reuben Peters, Brittany Petros, and Dale Poulter for their careful reading and thoughtful comments on this review prior to publication. I thank Timothy S. Oey for his assistance with the figures. Additionally, I thank the NIH for Grant GM56838 in support of my research on terpenoid cyclase structure and function.
David W. Christianson received the A.B. (1983), A.M. (1985), and Ph.D. (1987) degrees in chemistry from Harvard University and joined the faculty of the University of Pennsylvania in 1988. Currently the Roy and Diana Vagelos Professor in Chemistry and Chemical Biology, Christianson has been studying enzymes of terpenoid biosynthesis ever since he reported the first bacterial terpenoid cyclase structure 20 years ago.36 Christianson’s research has been recognized by several awards, including a Searle Scholar Award (1989), an Alfred P. Sloan Fellowship (1992), a Camille and Henry Dreyfus Teacher-Scholar Award (1993), the Pfizer Award in Enzyme Chemistry (1999), a Guggenheim Fellowship (2006), and the Repligen Award in Chemistry of Biological Processes (2013). Christianson was also Visiting Professor and Fellow in the Natural Sciences at Sidney Sussex College, University of Cambridge (2006), the Elizabeth S. and Richard M. Cashin Fellow at the Radcliffe Institute for Advanced Study at Harvard University (2015), and Visiting Professor of Chemistry and Chemical Biology at Harvard University (2016). An avid sailor, Christianson can often be found on the Chesapeake Bay and beyond as captain of the sloop Sea Symphony, berthed at home port Annapolis, Maryland.
The author declares no competing financial interest.