Enhancing structural diversity of terpenoids by multisubstrate terpene synthases

Introduction

Terpenoids constitute the largest class of natural products with more than 80000 known structures [1] and a broad range of bioactivities [2,3]. Despite their stunning diversity, all terpenes are biosynthetically derived from two general isomeric C₅ building blocks, dimethylallyl diphosphate (DMAPP, 1) and isopentenyl diphosphate (IPP, 2), via the mevalonate (MVA) or methylerythritol 4-phosphate (MEP) pathways. These two isomeric C₅ precursors are further condensed by prenyltransferases (PTs) in successive elongation reactions, resulting in geranyl diphosphate (GPP, 3), farnesyl diphosphate (FPP, 4), geranylgeranyl diphosphate (GGPP, 5), and longer prenyl diphosphates. The acyclic precursors are then converted into (poly)cyclic skeletons, including hemiterpenes, monoterpenes, sesquiterpenes, diterpenes, sesterterpenes, and triterpenes, by a large class of enzymes called terpene synthases (TSs) (Figure 1).

The reactions of TSs are one of the most important factors contributing to terpene diversity, as they often synthesize multiple products from a single substrate through complex cyclization cascades [4-10]. Based on the mechanism of initial carbocation generation, TSs generally fall into two main classes. Class I TSs generate an allylic cation from a prenyl substrate by depyrophosphorylation, whereas class II TSs utilize a general acid (a key Asp residue) to protonate the terminal C=C bond or epoxide group to yield a tertiary carbocation. The highly reactive carbocation is then converted to different carbocation intermediates, facilitated by the hydrophobic pocket of the TSs, which often results in multiple terpene products from a single prenyl substrate. In addition to their distinct mechanisms, the two major classes of TSs are classified according to their sequences, structures, and functions. For instance, class I TSs often have conserved sequence motifs, DDXXD and NSE/DTE, that bind trinuclear magnesium clusters for diphosphate abstraction, whereas class II TSs have a DXDD motif that acts as the catalytic acid. Recently, several novel unconventional TSs that share low sequence and structural similarities with classical TSs have been discovered and comprehensively reviewed [11,12].

In addition to the capability to generate multiple products using a single substrate, a growing number of TSs called multisubstrate terpene synthases (MSTSs) are capable of utilizing prenyl precursors with different chain lengths or configurations to synthesize diverse terpenoid products. Notably, MSTSs can also convert noncanonical prenyl substrates, including chemically synthesized analogs and bio-originated 6-, 7-, 8-, 11-, and 16-carbon substrates generated by methyltransferases or engineered lepidopteran mevalonate pathways. The multisubstrate features of these enzymes have often been characterized using in vitro assays. The in vivo activities of MSTSs were revealed by the development of an efficient precursor-providing chassis. The inherent features of MSTSs not only increase the structural diversity of terpenoids but also underscore their potential for generating new terpenoids through combinatorial biosynthesis. An important review published previously comprehensively addressed the transformation of synthetic prenyl-substrate analogs by TSs as well as TS-mimicking chemical transformations [13]. In this review, we discuss representative MSTSs originating from different species that use canonical prenyl substrates. We also highlight recent advances in the production of novel terpenoids by MSTSs using synthetic prenyl substrates. Finally, we focused on MSTSs that catalyze the transformation of naturally occurring noncanonical prenyl substrates.

Review

MSTSs using canonical prenyl diphosphate substrates

MSTSs from plants

Substrate-promiscuous TSs are widely spread in plants, which mainly produce linear terpenoids such as linalool (6), (E)-nerolidol (7) and (E,E)-α-farnesene (8) (Figure 2) [14,15]. Most plant MSTSs accept two prenyl substrates: C₅ and C₁₀ [16], C₁₀ and C₁₅ [17-20], and C₁₅ and C₂₀ [18]. For instance, PamTps1 from Plectranthus amboinicus (Lour.) Spreng has been characterized as bifunctional in converting compounds 3 and 4, respectively, to 6 and 7 both in vivo and in vitro (Table 1) [17]. In addition to the bifunctional plant TSs, a few plant MSTSs have been characterized using multiple prenyl substrates. Recently, four TSs belonging to the TPS-a subfamily [21] from common self-healing (Prunella vulgaris) were characterized in vitro to accept 3, neryl diphosphate (NPP), 4, (Z,Z)-FPP, 5, and/or nerylneryl diphosphate (NNPP) using purified recombinant enzymes (Table 1) [18]. A TPS-f subfamily enzyme CoTPS5 from Cananga odorata has been characterized to convert 3 to (E)-β-ocimene (9), 4 to 8, and 5 to diterpene α-springene (10) (Figure 2) [22]. Both in vitro assays and in vivo transgenic expression of CoTPS5 confirmed the absence of side products, indicating that CoTPS5 is highly selective for individual prenyl substrates and that the reaction is tightly regulated under different conditions. Notably, CoTPS5 was the first plant TS to produce 10 [22].

Although many MSTSs exhibit a broad substrate scope in vitro, their product profiles may be altered in vivo owing to the subcellular localization of enzymes and the availability of substrates in different intracellular compartments [23]. For instance, two nerolidol/linalool synthases from Antirrhinum majus (AmNES/LIS-1, -2) both synthesize 6 and 7 in vitro, but cytosol-localized AmNES/LIS-1 produces only 7, while plastid-localized AmNES/LIS-2 synthesizes 6 [24]. Similarly, in the case of CoTPS5, the transient expressed cytosol CoTPS5 in N. benthamiana only generated 8, while the plastid-localized CoTPS5 yielded 9 and 10 other than 8 (Table 1). These studies indicate that redirecting MSTSs to different subcellular compartments may facilitate the generation of multiple terpenoids in plants.

Recently, in addition to linear terpenoid-producing TSs, MSTSs that form cyclic terpenoids have been discovered in plants, further increasing our understanding of chemodiversity and biosynthesis of plant terpenoids. LcTPS2 from Leucosceptrum canum was characterized as a versatile TS that generated six macrocyclic sesterterpenoids (11–16) and two macrocyclic diterpenoids (17,18), representing the first macrocyclic terpenoids isolated from plants (Table 1, Figure 2) [19]. In addition to linear prenyl substrates, MSTSs can also accept partially cyclized substrates. A class I diterpene synthase SiTPS8 from Setaria italicais is capable of utilizing three copalyl pyrophosphate (CPP) stereoisomers that were generated by different class II TSs, including ent-CPP (19), (+)-CPP (20), and syn-CPP (21), to generate different diterpene skeletons 22–28, which were further converted to diterpenoids by a P450 monooxygenase (CYP99A17) (Table 1, Figure 2) [20]. These findings will enable further investigation of the functions of terpenoids in plants and crops.

MSTSs from fungi

Fungi are also prolific producers of terpenoids with diverse cyclic structures and important biological activities, which are of great interest. However, the number of known fungal MSTSs is currently limited, and researchers have focused on the promiscuity of their products rather than substrates [25-27]. Unlike plant MSTSs, fungal MSTSs convert natural substrates into cyclic skeletons. According to a phylogenetic tree constructed using 51 well-characterized class I TSs, clade III is of particular interest because most characterized di- and sester-TSs are enriched in this clade [28]. Two clade III TSs, FgMS and FgGS, from Fusarium graminearum J1-012 were characterized as promiscuous TSs with broad substrate specificities both in vitro and in vivo, indicating that TSs in clade III are more likely to be promiscuous. Using an efficient precursor-providing chassis, 50 terpenoids were generated via combinatorial biosynthesis using only two TSs and three PTs to generate 4, 5 or geranylfarnesyl diphosphate (GFPP, 29, Figure 1), representative products 30–33 are shown in Figure 3a [28]. Notably, FgMS is a chimeric enzyme (PTTS) consisting of an N-terminal class I TS domain and a C-terminal GFPP synthase domain. Therefore, to block the generation of 29, a variant of FgMS-D510A with an inactive PT domain, rather than wild-type FgMS, was used in combinatorial biosynthesis. Furthermore, critical residues controlling substrate specificity were identified using site-directed mutagenesis. Interestingly, when the aromatic residue Phe65 was replaced with Ala, the resulting variant F65A produced a novel 5/8/6/6 tetracyclic sesterterpene in the presence of 29 [28]. Domain swapping is another useful approach for changing the PTTS product profile. For example, EvVS from Emericella variecolor majorly produced diterpene variediene (34) with a minor production of sesterterpene (2E)-α-cericerene (33) in vitro (Figure 3a) [29]. By replacing the PT domain of EvVS with that of sester-TS EvSS, the resulting variant generated 33, which was not produced in vivo as the major product, both in vitro and in vivo. These studies revealed that altering and enhancing the supply of prenyl substrates can significantly change the product profile of promiscuous TSs, thereby generating terpenes with novel structures.

MSTSs from bacteria

According to previous studies, plants and fungi are the major producers of terpenoids [30-33]. Recently, an increasing number of TSs have been discovered in bacteria [32,33]. VenA from Streptomyces venezuelae ATCC 15439 was characterized as a promiscuous class I TS that converts 3 to geraniol (24.2% yield), 4 to seven sesquiterpenes (24.6% yield), and 5 to four diterpenes (31.2% yield), with venezuelaene A (35, Figure 3b) as the predominant product in vitro. Notably, compound 35 has an unprecedented 5/5/6/7 tetracyclic skeleton [34]. In addition to Actinomyces, MSTSs have been discovered in Bacillus. The large-TS BclTS from Bacillus clausii generated β-geranylfarnesene (36) and β-hexaprene (37) from 29 and hexaprenyl diphosphate (HexPP, 38, Figure 1), respectively. Similarly, a related TS, BalTS [35,36] from Bacillus alcalophilus was discovered to convert C₂₅, C₃₀, and C₃₅ prenyl diphosphates (39, Figure 1) into the corresponding β-prenes (36, 37, and 40, Figure 3b) [37]. Although BalTS shows no conserved motifs and distinct primary structures with class I TSs, its crystal structure reveals a similar overall structure of BalTS to the α-domain of class I TSs and therefore was proposed as class IB, a new subclass of TSs [37]. The discovery of TSs from bacteria not only expands the diversity of terpene skeletons but also the repertoire of TSs from nature.

TSs using noncanonical prenyl diphosphate substrates

Chemically synthesized noncanonical prenyl substrates

Noncanonical prenyl diphosphates are analogs of natural prenyl diphosphates. Most noncanonical prenyl diphosphate substrates are chemically synthesized. Classically, these prenyl analogs have been used as co-crystallization ligands [38], inhibitors of specific TSs [39], and tools to study the reaction mechanisms of cyclization cascades [40,41] which have been comprehensively addressed in important previous reviews [8,13]. Currently, noncanonical prenyl analogs have been synthesized to act as actual substrates of TSs to generate novel terpene skeletons, introduce reaction handles, and produce value-added compounds. A previous review has covered the advances of TS-catalyzed transformations of synthetic substrate analogs up to 2019 [13,42]. Here, we provide updated examples on this topic.

Recently, novel sesquiterpene backbones 41–44 were synthesized by feeding presilphiperfolan-8β-ol synthase (BcBOT2) with methyl-shifted FPP analogs 45–47 (Figure 4a) [43]. Three new homosesquiterpenes 48–50 were produced after the biotransformation of BcBOT2 with the FPP cyclopropylmethyl analog 51 (Figure 4b) [44]. With FPP ether derivative 52, pentalenene synthase (PenA) and the Δ⁶-protoilludene synthases (Omp6/7) from Omphalotus olearius several new tetrahydrofurano terpenoids 53–58 were obtained, some of them accompanied with pronounced olfactoric properties (Figure 4c) [45]. Notably, one tetrahydrofuranoterpenoid 59 is also formed as a major product in the BcBoT2 reaction, despite the low sequence similarity between these sesqui-TSs, which could be explained by similar active-site conformations to stabilize prenyl substrates. Similarly, limonene synthase (CLS) from Cannabis sativa and 5-epi-aristolochene synthase (TEAS) from Nicotiana tabacum were incubated with 11 synthetic prenyl analogs with ether, thioether, alkyne, or phenyl groups, and six of them (60–65) were converted into several novel monoterpenoids 66–71 [46] (Figure 5). In addition to regular chemically synthesized prenyl analogs, chemoenzymatic approaches have been developed to obtain new sesquiterpenes from prenols, providing an alternative approach for accessing FPP analogs [47]. These studies demonstrate the potential of TSs to utilize noncanonical synthetic prenyl analogs to yield unusual terpenoid skeletons and new value-added terpenoids.

In addition to expanding the repertoire of terpenoids, the biotransformation of noncanonical prenyl substrates by TSs provides insights into the mechanisms of cyclization reactions. β-Himachalene synthase (HcS) and (Z)-γ-bisabolene synthase (BbS) from Cryptosporangium arvum, and germacrene A synthase (SmTS6) from Streptomyces mobaraensis were chosen to convert four FPP analogs 72–75, which not only generated several new terpenoids (76–79), but also revealed the cyclization mechanisms of selected TSs [40] (Figure 6a). Similarly, two GGPP analogues 80 and 81 with shifted double bonds were synthesized to study the stereochemistry and cyclization mechanism of casbene synthase (CS) from the castor bean (Ricinus communis), which indicated a stereochemical course in accordance with the reported absolute configuration of casbene [41] (Figure 6b). The same GGPP isomers (80, 81) were employed to generate novel diterpene derivatives and revealed the cyclization mechanisms of 12 di-TSs [48]. Similarly, dihydro-GGPP (82) and dihydro-GFPP (83) have been synthesized for biotransformation using several di- and sester-TSs. The conversion of analogues 82 and 83 by TSs led to the production of ruptenes including compounds 84–90, which revealed the structure of the proposed intermediates for the cyclization reactions and therefore provided important insights into the reaction mechanism [49] (Figure 6c). With the aid of artificial prenyl analogs, a new route was developed to access a pool of unnatural terpenoids. It is worth noting that the rational design and synthesis of analogs play a valuable role in elucidating the cyclization mechanism of TSs, which further broadens our knowledge of the biosynthesis of terpenoids.

Naturally occurring noncanonical prenyl substrates

Most of the terpene biosynthesis is well defined by the ‘isoprene rule’ to form natural products by the polymerization of C₅ isoprene. Although terpenes with irregular carbon atoms (C₆, C₇, C₁₁, C₁₂, C₁₆, and C₁₇) have been characterized, they are thought to be synthesized by modifications after the formation of the terpene skeletons [50]. Recently, additional routes have been discovered for the production of noncanonical terpenoids, whose biosynthesis requires C-methyltransferases from bacteria. IPP/DMAPP methyltransferases have been shown to convert C₅ prenyl substrates to irregular C₆ (91–95), C₇ (96–100), and C₈ diphosphates (101), which could serve as building blocks for the generation of new terpenoids [51] (Figure 7). Furthermore, a series of noncanonical C₁₁, C₁₂, C₁₆, and C₁₇ prenyl substrates were synthesized in Escherichia coli harboring heterologously expressed IPP methyltransferase (IPPMT) from Streptomyces monomycini. Notably, polymethylated C₄₁, C₄₂, and C₄₃ carotenoids were produced by combining the endogenous terpene biosynthesis pathway and IPPMT, demonstrating the potential of this approach to expand the terpene structural space [52].

In addition to methylation of the elongation unit IPP, noncanonical prenyl substrates can also be prepared by modifying the prenyl substrate of TSs. For instance, the heterologous expression of GPP C2-methyltransferases with C₁₁-TSs and mevalonate biosynthesis enzymes in E. coli yielded 35 C₁₁ terpenes and 11 C₁₆ terpenes [53]. By introducing a GPP C2-methyltransferase from Pseudanabaena limnetica to yeast together with an engineered C₁₁-specific TS, 40 C₁₁ terpene scaffolds were produced, which significantly increased the chemical space of terpenoids [54]. More recently, the GPP C6-methyltransferase BezA was discovered in Streptomyces coelicolor [55] (Figure 8a). Further structure-based engineering of BezA successfully repurposed it to catalyze the unprecedented C6-methylation of FPP by a single residue substitution in its substrate-binding pocket [55]. Moreover, efforts have also been made to engineer the TSs to modulate their product selectivity with the noncanonical prenyl substrates. To enable the biotechnological synthesis of irregular terpenes, the product selectivity of 2-methylenebornane synthase from Pseudomonas fluorescenes was altered using a semi-rational engineering approach [56].

In contrast to GPP methylation, modification of FPP is catalyzed by the C-methyltransferase SpSodMT. In 2018, the biosynthesis of an unusual homosesquiterpene, sodorifen (102, Figure 1), from Serratia plymuthica 4RX13 was elucidated [57]. The in vitro and in vivo results revealed that a SAM-dependent-C-methyltransferase catalyzed methylation and cyclization reactions to form pre-sodorifen (103, Figure 8b), which was subsequently converted to 102 by TS [57]. Key residues lining the catalytic cavity of SpSodMT, Q57, F58, N219, V273, and L302, were found to affect product outcomes, and mutagenesis of these residues resulted in new C₁₆-prenyl substrates [58]. Selcted SpSodMT variants provided ten C₁₆ building blocks, including 103, plymuthenyl diphosphate (104), thorvaldsenyl diphosphate (105), weylandtenyl diphosphate (106), blixenyl diphosphate (107), kimlarsenyl diphosphate (108), serratinyl diphosphate (109), jacobsenyl diphosphate (110), hammershoyl diphosphate (111), and ancheryl diphosphate (112) (Figure 8c). Subsequently, terpentetriene synthase from Kitasatospora griseola (Cyc2) and kolavelool synthase from Herpetosiphon aurantiacus (HaKS) were identified as capable of converting these building blocks into C₁₆ scaffolds, whereas other selective sesqui-TSs failed to transform 104–112. After further modification with cytochrome P450 CYP720B1 in the yeast host, 28 noncanonical terpenoids were generated, 10 of them are shown in Figure 8c (113–122) [58]. Notably, the widespread biosynthesis of C₁₆ terpenoids was reported in a recent study in which biosynthetic gene clusters for C₁₆ terpenoids were identified and grouped into four types according to the number of MTs and TSs in the gene cluster [59]. A subset of methyltransferase genes was functionally characterized using engineered yeast, which has an enhanced supply of 4 (strain AM109) [60] and the main product of these enzymes was compound 103 [59]. Subsequently, 35 selective TSs in these gene clusters were characterized using a yeast chassis, and 47 noncanonical terpenoids were produced with 13 of them being elucidated (123–135, Figure 9), which enabled further studies on their functions and prompted the discovery of new types of terpenoids [59].

In addition to methylation of prenyl substrates, the lepidopteran mevalonate (LMVA) pathway can also form six-carbon homoisopentenyl pyrophosphate (HIPP, 136). The lepidopteran (butterflies and moths) pathway incorporates 3-ketovaleryl-CoA (137), instead of acetoacetyl-CoA (138), into the normal MVA pathway to generate compound 136 (Figure 10a). By heterologously expressing the LMVA pathway with a propionyl-CoA ligase and TSs, researchers have successfully produced several novel terpenes containing 16 carbon atoms in E. coli albeit at low titers [61]. To increase C₁₆ terpene titers, a later study redirected the 3-ketovaleryl-CoA formation step from the previous thiolase-dependent LMVA pathway to a β-oxidation LMVA pathway, and combined it with a promiscuous phosphatase, NudB, to produce C₆- and C₇-isoprenol. Notably, the final products, 3-ethyl-3-buten-1-ol (139) and 3-propyl-3-buten-1-ol (140), have potent fuel properties, highlighting the potential of this strategy for producing isoprenol analogs as next-generation biofuels (Figure 10b) [62].

Conclusion

During terpenoid biosynthesis, most TSs have strict substrate selectivity; nevertheless, some promiscuous TSs accept multiple prenyl substrates and produce various products. In nature, the biosynthesis of prenyl substrates may have subcellular locations, and the available types of prenyl substrates are limited, especially for noncanonical substrates in living cells. Therefore, the potential of TSs to generate terpenoids has been underestimated. With the development of synthetic biology technologies, an efficient precursor-providing chassis was constructed. Together with the accumulation of genome sequencing data, we systematically evaluated the function of TSs and discovered new terpenoids via genome mining. Nevertheless, for drug development, the accumulated terpene skeletons still require further functionalization, which requires additional genome-mining efforts for the discovery of tailored enzymes. Researchers have successfully expanded the chemical space of terpenoid biosynthesis using noncanonical prenyl substrates, which were synthesized using chemical approaches or via biosynthetic pathways. Many new terpenoids have been derived from chemically prepared prenyl analogs for decades, and until recently, the conversion of enzymatically modified noncanonical substrates has been utilized. New building blocks with irregular carbon numbers broaden the diversity of terpenoid structures. However, more systematic studies on noncanonical terpenoids are needed to study their biological activities.