Harnessing ortho-Quinone Methides in Natural Product Biosynthesis and Biocatalysis

Abstract

The implementation of ortho-quinone methide (o-QM) intermediates in complex molecule assembly represents a remarkably efficient strategy designed by Nature and utilized by synthetic chemists. o-QMs have been taken advantage of in biomimetic syntheses for decades, yet relatively few examples of o-QM-generating enzymes in natural product biosynthetic pathways have been reported. The biosynthetic enzymes that have been discovered thus far exhibit tremendous potential for biocatalytic applications, enabling the selective production of desirable compounds that are otherwise intractable or inherently difficult to achieve by traditional synthetic methods. Characterization of this biosynthetic machinery has the potential to shine a light on new enzymes capable of similar chemistry on diverse substrates, thus expanding our knowledge of Nature’s catalytic repertoire. The presently known o-QM-generating enzymes include flavin-dependent oxidases, hetero-Diels–Alderases, S-adenosyl-l-methionine-dependent pericyclases, and α-ketoglutarate-dependent nonheme iron enzymes. In this review, we discuss their diverse enzymatic mechanisms and potential as biocatalysts in constructing natural product molecules such as cannabinoids.

Graphical Abstract

INTRODUCTION

Natural product molecules are a profound source of inspiration for synthetic chemists and biochemists alike. The elaborate structures expertly assembled by Nature serve as blueprints for designing synthetic and biocatalytic routes to pharmaceuticals,¹ feedstock chemicals,² and other important materials.³ Understanding the fundamental mechanisms refined by Nature for the biosynthesis of natural products enables the development of improved methods for accessing complex molecules. Modern nucleic acid sequencing technologies have rapidly accelerated the discovery of the biosynthetic machinery responsible for the assembly of numerous natural products.^4,5 Such technological advancements have facilitated the study of natural product biosynthetic pathways from unculturable organisms,⁶ entire microbial communities,⁷ and higher order organisms such as macroalgae,⁸ plants,⁹ and animals.¹⁰ This influx of information has also resulted in the discovery of new enzyme classes and enzymes with unprecedented chemistry, accelerating the development of biocatalysts for chemical synthesis.

A prime example of a chemical motif designed by Nature and utilized by synthetic chemists is the ortho-quinone methide (o-QM), a reactive intermediate that has been extensively reviewed and often used in complex biomimetic syntheses and chemical manufacturing.^11–18 o-QMs are suspected to be involved in the formation of several bioactive natural products, for example, parvinaphthol C (1)¹⁹ and busseihydroquinone C (2)²⁰ isolated from plants and peniphenone D (3),²¹ (–)-xyloketal D (4),²² and epolone B (5)²³ derived from fungi (Figure 1a). In its simplest representation, an o-QM is composed of a 2,4-cyclohexadienone core with an exocyclic methylene ortho to the carbonyl group (Figure 1b). These innately reactive intermediates can also be depicted as zwitterionic or biradical species and may exist as E/Z isomers depending on steric hindrance from adjacent functional groups (Figure 1c). The reestablishment of aromaticity drives o-QM reactivity, and this logic also extends to nonbenzenoid aromatics, such as tropolones (Figure 1b), which favor aromaticity rather than o-QM dearomatization. In contrast to the well-documented para-quinone methide regioisomer (Figure 1b), o-QMs exhibit a greater charge dipole and are therefore less stable and more reactive such that nearly all o-QMs are unable to be isolated and will undergo rapid oligomerization in the absence of a suitable nucleophile.¹³ Further, the zwitterionic nature of o-QMs makes them both electrophilic and nucleophilic, rendering their utility as nucleophiles and electron-poor dienes particularly attractive for synthetic purposes.

Figure 1.

(a) Examples of natural products proposed to utilize a biosynthetic o-QM intermediate. o-QM portion of the molecules are highlighted in teal. (b) Representative structures of commonly observed quinone methides. (c) Canonical representations of ortho-quinone methide zwitterionic, geometric, and biradical isomers. (d) Common synthetic and biosynthetic approaches to o-QM formation and reaction outcomes.

The utility of o-QMs extends to a variety of complexity-building reactions, including cycloaddition, spirocyclization, and oligomerization reactions (Figure 1d). o-QMs are highly versatile yet challenging to employ due to rapid degradation and the formation of unwanted byproducts, impeding their selectivity and overall utility. In traditional synthetic approaches, careful consideration must be taken as to the reaction conditions used to generate the o-QM, especially when invoked at later stages in the route. Figure 1d summarizes some of the strategies for o-QM initiation, including examples of the most commonly used o-QM precursors. Many of these strategies require phenolic protection and activation of the benzylic carbon prior to thermal, photolytic, or chemical generation of the o-QM and exorbitant amounts of a nucleophile or diene to mitigate oligomerization (Figure 1d). Despite the versatility of the o-QM intermediate, such reaction requirements undermine its accessibility in synthetic endeavors. As a complementary approach, biosynthetic pathways involving enzymatic o-QM intermediates have the potential to expedite the development of biocatalytic methods that are more efficient, selective, and sustainable relative to traditional synthetic techniques.

Most of the evidence for o-QM intermediates in natural product biosynthesis has relied on biomimetic syntheses to replicate crucial C–C or C–O bond forming steps via o-QMs.^12,15,16 The biosyntheses of many natural products perceived to be derived from o-QM intermediates, including many acylphloroglucinol-derived natural products and related meroterpenoids, have yet to be fully understood.^24–31 However, recent investigations have uncovered a diverse collection of enzymes from bacterial, fungal, and plant sources capable of generating and directing regio-, stereo-, and chemoselective reactions via o-QM intermediates. Perhaps even more remarkable is the fact that these enzymatic reactions take place under aqueous conditions traditionally avoided due to the nucleophilicity of water promoting rapid quenching of o-QMs. Given these unique attributes, enzymes capable of generating o-QMs are well positioned to supplement synthetic endeavors that require these highly reactive intermediates. Herein, we review the known enzymes associated with o-QM generation in natural product biosynthetic pathways, including examples of berberine bridge enzyme-like (BBE-like) oxidases, tropolonic hetero-Diels–Alderases, S-adenosyl-l-methionine (SAM)-dependent pericyclases, and α-ketoglutarate-dependent nonheme iron oxygenases. In addition to mechanistic details, the biocatalytic potentials of these enzymes are also discussed including substrate scope, scalability, and use in chemoenzymatic syntheses. Finally, an outlook on the discovery of new o-QM-generating enzymes with biocatalytic potential is presented.

BERBERINE BRIDGE ENZYME-LIKE OXIDASES

Flavin-dependent enzymes are versatile redox catalysts in primary and secondary metabolic pathways and account for more than half of the characterized enzymes involved in generating o-QMs.^32,33 This versatility is exemplified not only by the diversity of substrates accepted but also in the impressive chemo-, regio-, and stereoselectivity displayed by the enzymes. Coincidentally, all of the characterized flavin-dependent enzymes associated with o-QM formation belong to the BBE-like oxidases family.³⁴ The BBE-like oxidase name is derived from the first characterized enzyme example, (S)-reticuline oxidase (EcBBE) from the California poppy (Eschscholzia californica), which catalyzes formation of the intramolecular C–C bond known as the “berberine bridge” observed in many benzylisoquinoline alkaloids.^35–37 BBE-like oxidases can facilitate many types of oxidation reactions, including intermolecular C–C bond formation, cyclizations, nucleophilic additions, and dehydrogenation reactions.³⁴ Structurally, BBE-like oxidases share a fold comparable to that observed in vanillyl-alcohol oxidases.^38–40 These enzymes are distinguished by the bicovalent attachment of flavin adenine dinucleotide (FAD) to the protein by histidine and cysteine residues at the 8α- and 6-positions of FAD, respectively (Figure 2).^41,42 This bicovalent attachment increases the redox potential of FAD, provides additional structural stability to the active site, and prevents cofactor dissociation.^43–46 Several BBE-like oxidases are hypothesized to invoke formation of an o-QM intermediate.^47–49 In these particular examples, the oxidized FAD cofactor facilitates hydride abstraction from the benzylic carbon and molecular oxygen acts as the terminal electron acceptor to recycle the reduced FAD, generating hydrogen peroxide as the sole byproduct in the catalytic cycle.

Figure 2.

Crystal structure of EcBBE (PDB 3D2H) illustrating bicovalent attachment of His104 and Cys166 to the 8α- and 6-positions of the isoalloxazine ring of FAD.

BBE-like oxidases are most well-recognized for their roles in the biosynthesis of phytocannabinoids (more commonly known as cannabinoids), natural products consisting of an isoprenylated resorcinol core with a para-positioned alkyl side chain. Cannabinoids are predominantly found in Cannabis sativa L. (Cannabaceae), a plant well-known for its recreational and medicinal applications, and have also been isolated from certain species of liverworts and fungi.^50–53 To date, over 100 different cannabinoids have been isolated and characterized.⁵⁴ Three primary constituents, cannabichromenic acid (CBCA, 10), tetrahydrocannibinolic acid (THCA, 11), and cannabidiolic acid (CBDA, 12), are derived from the same biosynthetic precursor, cannabigerolic acid (CBGA, 7) (Figure 3). The decarboxylated forms of CBCA (10), THCA (11), and CBDA (12), cannabichromene (CBC), tetrahydrocannabinol (THC), and cannabidiol (CBD), respectively, are among the most commonly recognized cannabinoids. Hexanoic acid (6) is converted to CBGA (7) through a series of enzymatic reactions requiring an acyl activating enzyme (AAE), tetraketide synthase (TKS), olivetolic acid cyclase (OAC), and aromatic prenyltransferase (ArPT). Rigorous biochemical investigations have elucidated the function and relationship of three BBE-like oxidases responsible for catalyzing the oxidative cyclizations of CBGA (7) into their respective acid products: THCA synthase, CBDA synthase, and CBCA synthase (Figure 4).^55–57 All three enzymes facilitate hydride abstraction from the benzylic position of CBGA (7) by the N-5 position of FAD and are differentiated by their ability to uniquely position CBGA (7) in the active site, resulting in different cyclized products.

Figure 3.

Abbreviated biosynthesis of cannabinoids. Structural diversification is generated by BBE-like enzymes CBDA synthase, THCA synthase, and CBDA synthase. Abbreviations: AAE, acyl activating enzyme; TKS, tetraketide synthase; OAC, olivetolic acid cyclase; ArPT, aromatic prenyltransferase.

Figure 4.

Proposed mechanisms for (a) THCA synthase, (b) CBDA synthase, and (c) CBCA synthase cyclization via o-QM intermediates.

There are two competing mechanisms for the subsequent cyclization reaction resulting in a carbocation shift or o-QM formation. Although a carbocation intermediate (8) is the more commonly depicted in the literature for THCA and CBDA cyclization, this zwitterionic species is simply a resonance structure of the o-QM (9) that is generated upon phenol deprotonation. Elucidation of the formal mechanism for CBCA (10), THCA (11), and CBDA (12) cyclization by each respective BBE-like oxidase has yet to be resolved, but proposed mechanisms are shown in Figure 4. In general, an active site tyrosine residue is proposed to deprotonate the C-5 hydroxy group of CBGA (7), initiating dearomatization of the benzene ring and hydride abstraction by the oxidized FAD cofactor (FAD_ox) to generate an o-QM intermediate and reduced FAD (FAD_red). Intramolecular cyclization takes place along with activation of oxygen and elimination of hydrogen peroxide to regenerate FAD_ox and the cyclized product (Figure 4).

THCA synthase has received the most attention of all reported cannabinoid BBE-like oxidases. The first reports identified a 74 kDa monomeric protein from C. sativa leaf extracts capable of converting CBGA (7) to THCA (11).⁵⁵ Subsequent sequencing efforts identified a 1635 base pair (bp) open reading frame encoding THCA synthase, which was heterologously expressed in insect cells using a baculovirus expression system.⁵⁸ The tertiary structure of THCA synthase was determined to 2.75 Å resolution by X-ray crystallography, revealing key amino acid residues important for substrate orientation and stabilization.⁴⁹ In accordance with previously characterized BBE-like oxidases, Cys176 and His114 were covalently tethered to the 6- and 8α-positions of FAD, respectively (Figure 5a, green). Ten additional hydrogen-bonding interactions and a disulfide bridge across Cys37 and Cys99 stabilize FAD within THCA synthase and provide structural integrity to the active site.

Figure 5.

Protein homology models comparing key active site residues of BBE-like cannabinoid synthases. (a) Overlay of THCA synthase (PDB 3VTE, green), CBDA synthase model (blue), and CBCA synthase model (orange). (b) Overlay of CBCA synthase model (orange) with DCA synthase model (pink). Residue numbering is relative to THCA synthase numbering.

Mutagenesis experiments have revealed several key active site residues essential for enzyme activity.⁴⁹ In particular, H114A and T484F completely abolished activity, whereas H292A decreased activity by 95%. The loss of activity in the H114A variant can be rationalized by the loss of covalent attachment to FAD, affecting the positioning and redox potential of the cofactor. Tyr484 is hypothesized to facilitate deprotonation of the C-5 phenol, a necessary step for generating the o-QM and promoting regioselective cyclization (see Figure 4). For reasons not fully understood, the carboxylic acid moiety in the substrate is essential for enzymatic activity.

In vitro experiments with fiber-type C. sativa (hemp) led to the discovery of a second BBE-like oxidase, CBDA synthase.⁵⁶ Using a similar homology-based cloning strategy to that previously implemented for THCA synthase, a 1632 bp open reading frame for CBDA synthase was identified, sharing ~84% sequence identity with THCA synthase.⁵⁹ Sequence analysis confirmed the presence of the RSGGH and CXXI/V/LG motifs consistent with bicovalent attachment to FAD. Biochemical characterization revealed that CBDA synthase catalyzed a mechanistically similar oxidative cyclization reaction to THCA synthase but generated a different cyclized product. Following intramolecular cyclization post-o-QM formation, a second basic residue in CBDA synthase is hypothesized to deprotonate a terminal carbon of the geranyl side chain of CBGA (7) to quench the carbocation intermediate (Figure 4b). This deprotonation step and substrate positioning in the active site are anticipated to drive the divergent reaction outcomes between THCA synthase and CBDA synthase. Protein homology models comparing the published crystal structure of THCA synthase with CBDA synthase revealed only a small number of differences, with high conservation of the key active site residues responsible for THCA synthase activity (Figure 5a, blue). Recent investigations into the structure–function relationships of THCA synthase and CBDA synthase revealed that most single-point THCA synthase substitutions generated to reflect the corresponding residue in CBDA synthase had little effect on activity or product specificity.⁶⁰ However, a CBDA synthase A414V variant created to reflect the homologous THCA synthase residue exhibited a 3.3-fold increase in CBDA (12) production and a 19-fold increase in THCA (11) production.⁶⁰ Until a crystal structure of CBDA synthase is elucidated, further mutagenesis studies exploring the structure–function relationship between these two enzymes will provide further insight into the mechanisms of product differentiation.

A third BBE-like oxidase, CBCA synthase, has been characterized from C. sativa and was the first plant chromene-forming oxidase identified. Chromene is a common structural motif observed in natural products, including several compounds with significant antimicrobial and cytotoxic properties.^57,61–63 A similar mechanistic initiation as observed with THCA and CBDA synthases can be envisioned for CBCA synthase by formation of the o-QM intermediate. CBCA synthase then facilitates nucleophilic attack by the phenolic oxygen, similar to THCA synthase, but differs by generating a mixture of enantiomeric chromenes (Figure 4c). The reaction is known to generate a 5:1 scalemic mixture, but the assignment of absolute configuration as to which is the favored remains unknown.⁶⁴ Similar to CBCA synthase, protein homology models do not indicate any significant variations within the active site (Figure 5a, orange). The difference is the stereoselectivity observed between THCA synthase and CBDA synthase, which produce one enantiomeric product exclusively, while the mixture of enantiomers generated by CBCA synthase suggests that the features governing cyclization selectivity may differ from the other two aforementioned cannabinoid synthases.

A CBCA synthase homologue was the only reported FAD-dependent oxidase capable of generating chromenes until the recent discovery of daurichromenic acid synthase, which shares 49% sequence identity with THCA synthase.^65,66 Daurichromenic acid (DCA, 15), isolated from Rhododendron dauricum (Ericaceae), is a meroterpenoid natural product structurally related to cannabinoids with potent anti-HIV properties.⁶⁷ DCA (15) and its biosynthetic precursor, grifolic acid (13), differ from the previously described cannabinoids by the longer, farnesyl group and shorter alkyl chain (Figure 6a). DCA synthase generates the chromene moiety via a similar o-QM intermediate to CBCA synthase (14) but differs in that a single enantiomer is produced (DCA, 15), suggesting that cyclization may occur within the active site (Figure 5b). Further experiments demonstrated that DCA synthase can accept grifolic acid derivatives (16, 18) with prenyl groups of various lengths but exhibited no activity with CBGA (7). This is presumably due to substrate binding interference by the pentyl chain, as DCA synthase was able to react with cannabigerorcinic acid (18), which only differs from CBGA (7) by alkyl chain length (Figure 6b). DCA synthase also showed no activity with the derivative lacking the carboxylic acid (grifolin, 20), which is consistent with the carboxylic acid moiety requirement observed with THCA, CBDA, and CBCA synthases.

Figure 6.

In vitro reactions with DCA synthase. (a) Confirmation of the role of DCA synthase in daurichromenic acid (15) biosynthesis. (b) Substrate screening with DCA synthase exhibits activity with various prenyl attachments but no activity with grifolin (20) or cannabigerolic acid (7). (c) Structurally related cannabinoids isolated from liverworts and fungi.

Beyond Cannabis and Rhododendron species, cannabinoids have been isolated from several species of lower plants and fungi. Two CBCA-like natural products, cannabiorcichromenic acid (21) and its halogenated analogue 8-chlorocannabiorcichromenic acid (22), have been isolated from the fungus Cylindrocarpone olidum (Nectriaceae).⁶⁸ Albatrellus spp. are known to produce confluentin (23), the decarboxylated analogue of DCA (15).⁶⁹ Several bibenzylic cannabinoids with THCA-like cyclizations have been isolated from the liverworts Radula perrottetii, R. marginata, and R. laxiramea, including (–)-cis-perrottetinene (24) and perrottetinenic acid (25) (Figure 6c).^51–53 Given the structural similarities to previously studied cannabinoids, it is likely that the enzymes responsible for oxidative cyclization in these bibenzylic natural products also belong to the FAD-dependent subfamily of BBE-like oxidases. Identification and characterization of these enzymes will create opportunities to engineer unprecedented cannabinoid derivatives with unique prenyl moieties, alkyl chains, and cyclized patterns. Additionally, the presently characterized cannabinoid synthases could be used as genetic hooks in attempting to identify homologues from publicly available genomic data sets that may accept similar substrates and perform the desired chemistry on complementary substrate scopes.

Bacterial flavoenzymes are well-documented to catalyze a broad range of redox reactions in natural product biosynthesis.^32,33 Coincidentally, the only two reported bacterial enzymes capable of generating o-QMs also belong to the BBE-like oxidase family of flavoenzymes. Next-generation sequencing of the marine bacterium Streptomyces sp. CNH-287 revealed a biosynthetic gene cluster (BGC) putatively capable of producing the tetrachlorinated alkaloid (–)-chlorizidine A (28, clz).⁴⁷ Genetic knockout experiments and in vitro biochemical characterization of the putative BBE-like oxidase Clz9 confirmed its role in chlorizidine A (28) biosynthesis. This reaction is thought to be mechanistically similar to cannabinoid synthases, initiated by benzylic hydride abstraction of prechlorizidine (26) via a bicovalently tethered FAD cofactor and phenolic deprotonation by a basic residue within the active site to generate the reactive o-QM intermediate 27. Clz9 catalyzes a stereoselective, intramolecular cyclization by nucleophilic addition of the pyrrole nitrogen, ultimately generating the unusual dihydropyrrolizine ring of chlorizidine A (28, Figure 7a).

Figure 7.

(a) Proposed mechanisms for cyclization vs dehydrogenation with microbial BBE-like oxidases Clz9 and Tcz9. (b) Comparing and contrasting catalytic functions of Clz9 and Tcz9 with non-native biosynthetic precursors.

Support for the proposed benzylic functionalization via an o-QM in Clz9 was provided with the recent discovery of a second bacterial BBE-like oxidase, Tcz9. Next-generation sequencing of a taxonomically distinct marine Actinomycete strain AJS-327 revealed a BGC with striking similarities to the BGC associated with (–)-chlorizidine A (28) biosynthesis and was proposed to be linked to the production of two novel tetrachlorinated alkaloids produced by the strain, dihydrotetrachlorizine (29) and tetrachlorizine (31).⁴⁸ Genes in the associated cluster, abbreviated tcz, were heterologously expressed. In vitro biochemical characterization of the BBE-like oxidase from the BGC, Tcz9, confirmed its catalytic function acting as a dehydrogenase in tetrachlorizine (31) biosynthesis (Figure 7a). Similar to Clz9, an o-QM intermediate (30) is proposed from hydride abstraction on dihydrotetrachlorizine (29). Rather than undergoing intramolecular cyclization with the pyrrole, the o-QM intermediate is deprotonated to yield the dehydrogenated tetrachlorizine (31) product. The C-15 carbonyl of dihydrotetrachlorizine (29) is thought to influence the reaction outcome by lowering the pK_a of the α-proton, but further substrate screening is required to provide evidence for this hypothesis. This is the first reported example of a dehydrogenated product generated via an o-QM intermediate, further expanding the utility of o-QMs and the unique chemoselectivity facilitated by enzymes.

Clz9 and Tcz9 have also been shown to catalyze intriguing oxidative reactions with non-native substrates (Figure 7b).⁴⁸ Incubation of Clz9 with dihydrotetrachlorizine (29), the biosynthetic precursor to tetrachlorizine (31), yielded a mixture of dehydrogenated (31) and cyclized (32) products. This indicates Clz9 preferentially acts as a cyclase but must also compete with dehydrogenation in the presence of the C-15 carbonyl moiety. More intriguing are the products generated upon incubation of Tcz9 with prechlorizidine (26), the biosynthetic precursor to (–)-chlorizidine A (28). Not only is Tcz9 capable of acting as a cyclase, but it also performs two subsequent dehydrogenation reactions on prechlorizidine (26), generating two isolable o-QM configurational isomers that are stable at room temperature (33, 34). These products provide key mechanistic insight into how these BBE-like oxidases function and demonstrate their exceptional utility in generating a variety of oxidized products. Further structure–activity relationship studies will reveal key mutations that are responsible for the functional differences between these two enzymes.

TROPOLONIC HETERO-DIELS–ALDERASES

The Diels–Alder reaction is one of the most powerful concerted pericyclic transformations for efficiently constructing complex natural product scaffolds. Although Diels–Alder reactions have been utilized in organic synthesis for nearly a century and have long been postulated as key steps in biosynthetic reactions, Diels–Alderases belong to a relatively new enzyme family.^18,70–73 Several multifunctional enzymes have demonstrated the ability to catalyze concerted [4+2] cycloaddition reactions, but the discovery of SpnF in spinosyn A biosynthesis ushered in a wave of standalone intramolecular Diels–Alder cyclases.⁷⁴ Recently, the first standalone [4+2] intermolecular carbocyclase, MaDA, was characterized from Moraceae alba and exhibits promising biocatalytic utility.⁷⁵

Hetero-Diels–Alderases have also received considerable interest for their suspected role in meroterpenoid biosyntheses.^76,77 Many of these proposed biosyntheses suggest an o-QM acts as an electron-deficient diene that reacts with a dienophile, such as an alkene, forming a dihydropyran ring. This motif is common in hundreds of plant meroterpenoids, especially those found in Eucalyptus spp. and Psidium guajava L. (guava).^{28–31,78–80} Synthetic approaches have demonstrated some of these hetero-Diels–Alder (hDA) reactions can occur nonenzymatically, but one class of meroterpenoid natural products, tropolonic sesquiterpenoids, has proven difficult to synthesize biomimetically.^81,82 Furthermore, the isolation of enantiopure metabolites suggests enzymatic influence, as a nonenzymatic reaction would theoretically generate a racemic mixture. Given the therapeutic interests and structural complexities of example compounds shown in Figure 8 (5, 35–37), identifying enzymes that can catalyze stereospecific hDA reactions using a reactive species such as a tropolone o-QM will likely have tremendous biocatalytic value.

Figure 8.

Representative examples of tropolonic sesquiterpenoids likely derived from hetero-Diels–Alder cyclization reactions. The o-QM portion is highlighted in teal.

Genes associated with the biosynthesis of humulene (44) and the tropolone core have been previously characterized and are highly conserved across fungal species known to produce tropolonic sesquiterpenoids.^83,84 However, the key linkage between the aromatic core and terpene remained elusive. The first tropolonic hetero-Diels–Alderase was reported in 2018 with the identification of a gene cluster in Sarocladium schorii (previously Acremonium strictum IMI 501407) that resembled previously characterized gene clusters associated with the production of tropolones, but also included several genes with unknown functions.⁷⁶ Heterologous expression of the captured gene cluster in Aspergillus oryzae led to the production of xenovulene A (46), a meroterpenoid natural product and potent antagonist for the human γ-aminobutyrate A (GABA_A) benzodiazepine receptor with promising antidepressant properties (Figure 9).

Figure 9.

Proposed biosynthesis of xenovulene A (46) via a tropolonic o-QM intermediate. The function of AsR5 was not verified in vitro; however, the presence of shunt products 47–50 is evidence supporting an o-QM intermediate.

In xenovulene A (46) biosynthesis, the humulene (44) constituent is produced by AsR6 from farnesyl diphosphate (43), while stipitaldehyde (40) is generated in a two-enzyme cascade starting with aldehyde 38 to form o-QM 39, which undergoes ring expansion via AsL3. The two pieces, stipitaldehyde (40) and humulene (44), were thought to be coupled by AsR5, a 401 amino acid protein with no significant sequence homology to any reported Diels–Alderases, as indicated by knockout experiments. Unfortunately, in vitro experiments investigating the function of AsR5 were inconclusive, as AsR5 exhibited no activity when attempting to couple stipitaldehyde (40) with humulene (44). It is proposed that 40 requires further oxidation by a cytochrome P450 embedded in the BGC (AsR2), forming hemiacetal 41 prior to hDA cyclization. This intermediate could not be isolated, but nonenzymatic dehydration and formation of a tropolonic o-QM could explain the isolable shunt products 47–50. Further evidence for enzymatic o-QM formation and hDA cyclization is supported by the isolation of a single stereoenriched product, 45, and biomimetic syntheses that have demonstrated hDA reactions between a tropolonic o-QM and humulene (44) require harsh reaction conditions,^81,82 suggesting the reaction is unlikely to occur spontaneously.

Shortly after the characterization of the xenovulene A (46) biosynthetic gene cluster, a second gene cluster associated with the production of the potent anti-glioma meroterpenoid eupenifeldin (36) was identified by genomic analysis and genetic disruption of Phoma sp. CGMCC 10481.⁷⁷ Sequence analysis confirmed genes with high sequence identity to those previously reported in tropolone and humulene (44) biosynthesis, including a gene encoding a putative hetero-Diels–Alderase (eupF). In vitro experiments confirming the function of the putative hetero-Diels–Alderase were inconclusive yet again due to protein solubility issues and the instability of the tropolone o-QM precursor (Figure 10). Later the same year, whole genome sequencing of Penicillium janthinellum led to the discovery of a third biosynthetic gene cluster (eupfA–J) containing all the necessary genes from tropolonic sesquiterpenoid production, including another putative hetero-Diels–Alderase (EupfF) with 67% identity to EupF.⁸⁵ A key finding in this work was the in vitro characterization of EupfE, a short-chain dehydrogenase that reduces stipitaldehyde (40), this time generated in a three-enzyme cascade from aldehyde 51, to the alcohol 52, priming this intermediate for dehydration and subsequent o-QM formation. EupfF exhibited similar solubility issues to previous accounts, but follow-up attempts to obtain the previously reported homologue EupF as a soluble protein eventually proved to be successful. Alcohol 52 can undergo spontaneous dehydration, o-QM formation (53), and nonenzymatic cyclization with 55 to generate 57, but the addition of EupfF significantly accelerates the reaction and generates a single diastereomer product, neosetophomone B (35). However, EupfF did not exhibit any activity to catalyze a second hDA reaction, leaving the possibility open for a second hetero-Diels–Alderase enzyme to complete the biosynthesis of bistropolonic meroterpenoid eupenifeldin (36). The efficient construction of stereospecific products via o-QMs using prochiral substrates has tremendous biocatalytic utility. Although this family of intermolecular hetero-Diels–Alderases is relatively new, genome mining efforts and biochemical validation will unveil new enzymes with similar capabilities.

Figure 10.

Proposed biosynthesis of eupenifeldin (36) via two o-QM intermediates. Genes responsible for production of neosetophomone B (35), EupfA–F, and EupF, a homologue of EupfF, were characterized by heterologous expression or in vitro experiments.

SAM-DEPENDENT O-METHYLTRANSFERASE-LIKE PERICYCLASE

SAM-dependent methyltransferases are a ubiquitous family of enzymes in primary and secondary metabolism. Canonical SAM-dependent methyltransferases catalyze the transfer of a methyl group to C, N, O, or S atoms from SAM, but this family of enzymes has expanded over time to include several nonmethylation reactions.^86–89 LepI is one such example of a noncanonical SAM-dependent O-methyltransferase, acting as a dehydratase and pericyclase in leporin C (65) biosynthesis.⁹⁰ Pathway reconstruction of key genes associated with leporin C (65) biosynthesis unveiled LepI is not essential for production; however, in vitro experiments confirmed that LepI is required for the accelerated and exclusive production of leporin C (65). The pathway begins in a three-enzyme cascade from l-phenylalanine (58) to yield 59. Reduction of 59 and production of the alcohol intermediate 60 by LepF leads to nonenzymatic dehydration and formation of both (E/Z) isomers of the o-QM intermediate (61), yielding multiple intramolecular and inverse electron demand hetero-Diels–Alder reaction outcomes (62–64) (Figure 11). In comparison, incubation of 60 with LepI leads to the stereospecific production of leporin C (65) via an o-QM. Small amounts of endo-64 were detectable at early time points when 60 was incubated with LepI; however, LepI can recycle this byproduct via a retro-Claisen rearrangement to produce leporin C (65) exclusively. Crystallographic analysis of LepI has revealed key active site residues that form a hydrogen-bonding network for proper substrate stabilization and orientation. Although the precise catalytic role of the SAM cofactor is not fully understood, the positively charged sulfonium ion is suggested to electrostatically stabilize o-QM intermediate (E)-61.⁹¹ Understanding the role of the SAM cofactor will be critical in genome mining and protein engineering efforts to broaden the utility of the SAM-dependent methyltransferase enzyme family capable of manipulating o-QMs for additional oxidative reactions.

Figure 11.

Leporin C (65) biosynthetic pathway. In the absence of LepI, 60 undergoes spontaneous dehydration and forms a mixture of nonenzymatic intramolecular Diels–Alder (IMDA) and hetero-Diels–Alder (hDA) products. Structures in gray are generated nonenzymatically. LepI can recycle endo-IMDA product 64 to stereoselectively produce leporin C (65).

α-KG-DEPENDENT NONHEME IRON OXYGENASES

α-Ketoglutarate (α-KG)-dependent nonheme iron oxygenases are a versatile family of enzymes known to catalyze remarkably diverse reactions, including hydroxylation, halogenation, epoxidation, desaturation, epimerization, endoperoxidation, ring contraction, and ring expansion.^92–94 As their name suggests, these enzymes require α-KG for activation of a Fe(II) cofactor. ClaD is an α-KG-dependent nonheme iron oxygenase associated with the biosynthesis of penilactone and peniphenone natural products produced by Penicillium spp., as confirmed by genetic knockout experiments.⁹⁵ Penilactones and pheniphenones are aromatic polyketide fungal metabolites derived from clavatol (77) coupled with a range of adducts, giving rise to diverse biological activities (3, 66–76, Figure 12a).^21,96–98 ClaD shares high sequence identity (54%) with CitB, an α-KG-dependent nonheme iron oxygenase known to catalyze a benzylic hydroxylation in citrinin biosynthesis. In vitro experiments demonstrated ClaD also hydroxylates clavatol (77) regioselectively, activating this biosynthetic intermediate for o-QM formation through hydroxyclavatol (78) to yield 79 (Figure 12b).⁹⁵ This distinguishes ClaD from the previously described BBE-like oxidases, tropolonic hetero-Diels–Alderases, and SAM-dependent O-methyltransferase LepI, which catalyze “direct” o-QM formation, stabilizing and orienting the o-QM within the active site for further manipulation. Instead, ClaD facilitates “indirect” o-QM formation, activating clavatol (77) for subsequent nonenzymatic dehydration. This process can be observed by isotope labeling experiments with H₂¹⁸O, which can be incorporated by reversible dehydration of hydroxyclavatol (78) and nucleophilic addition by isotope-labeled water.⁹⁵ The reversibility of the dehydration and nucleophilic addition reactions via the o-QM allows other nucleophiles present in solution to react, as observed by the coupling of various electron-rich functional groups with clavatol (77). This benzylic functionalization strategy developed by Nature has been validated by biomimetic syntheses of several clavatol-derived natural products via o-QMs.⁹⁹

Figure 12.

(a) Clavatol (77)-derived natural products from Penicillium sp. (b) Regioselective benzylic hydroxylation by α-KG-dependent nonheme iron oxygenase ClaD, which undergoes “indirect” (spontaneous) dehydration to produce a reactive o-QM.

BIOCATALYTIC UTILITY OF ENZYMATIC o-QMS

Synthetic approaches to o-QM formation have been reported in numerous biomimetic natural product syntheses, but these approaches typically suffer from poor atom economy and require toxic chemical reagents, only to achieve the same reaction outcome that Nature has already optimized. Several complex molecules derived from o-QMs require Herculean efforts to achieve robust yet concise processes that can rival the elegance of Nature;¹⁰⁰ in other cases, the structural complexity designates these molecules to be synthetically intractable. Thus, leveraging enzymes for biocatalytic generation of o-QMs and engineering these proteins to perform chemo-, regio-, and stereoselective reactions efficiently will supplement traditional synthetic approaches.

Heterologous expression is an attractive solution to access rare or unnatural cannabinoids that cannot be efficiently produced by current cultivation and synthetic methods.^101–103 THCA synthase, CBDA synthase, and CBCA synthase have received considerable attention for their ability to catalyze cyclization reactions and have been successfully integrated into numerous eukaryotic expression systems.^{102,104–107} Mutagenesis experiments have identified several key active site residues, but additional mutagenic and crystallographic experiments are required to fully understand how these enzymes catalyze different cyclization reactions from the same o-QM intermediate. The recent characterization of DCA synthase^65,66 enables further opportunities to study the substrate promiscuity and enantioselectivities of BBE-like cannabinoid synthases. Further genome mining and engineering efforts of BBE-like enzymes across all domains of life may provide access to additional cannabinoid-like scaffolds.

Synthetic approaches coupling clavatol-like precursors with nonbiogenic coupling partners have been reported; however, these typically require high temperatures, organic solvents, and protection of the clavatol (77) intermediate prior to o-QM generation. Alternatively, biological systems have been developed to achieve this same purpose using milder conditions and aqueous buffers. Cloning the fungal nonreducing polyketide synthase (PKS) gene pksCH-2 into an Aspergillus oryzae heterologous host produced chaetophenol A (80) and derivatives 81 and 82. These intermediates are reductively cyclized by an endogenous enzyme in vivo, yielding the isochromenes 83–85 that tautomerize to o-QMs (86–88), leading to production of chaetophenol E (93) and new oligomeric analogues.¹⁰⁸ Acid- or base-promoted tautomerization or chemical oxidation are also invoked for o-QM formation. In total, 40 novel polyketide oligomers (e.g., 92 and 93), azaphilone-type molecules (e.g., 89–91), and indole-polyketide hybrid molecules (e.g., 94–96) were prepared using this diversity-oriented semisynthetic process and screened for antiviral activity (Figure 13a). A similar combinatorial approach was developed by incubating nucleophiles with the hydroxyclavatol (78)-producing strain Penicillium crustosum PRB-2, which led to the isolation of 15 novel clavatol-alkaloid derivates screened for antiviral activity (e.g., 97–100, Figure 13b).¹⁰⁹ These semisynthetic approaches achieve both efficient construction of o-QMs and structural diversification for rapid combinatorial library generation.

Figure 13.

Biocatalytic developments for in vivo o-QM formation and benzylic functionalization with non-natural substrates, resulting in the rapid generation of diverse “pseudo-natural product” libraries. (a) Chaetophenol A (80) undergoes reductive cyclization by an endogenous enzyme in the heterologous host, which readily produces a highly reactive o-QM intermediate via a retro-Michael reaction. PKS enzyme domain abbreviations: ACP, acyl carrier protein; AT, acyl transferase; KS, ketoacyl synthase; MT, methyltransferase; PT, product template; R, reductase; SAT, starter unit:ACP transacylase. (b) Production of clavatol-like alkaloids by reacting hydroxyclavatol (78) with indoles and aniline-type nucleophiles in vivo.

ClaD and CitB have also demonstrated remarkable substrate promiscuity to generate benzylic alcohols that can undergo facile o-QM formation.¹¹⁰ In total, 21 of the 22 aromatic substrates screened in vitro were accepted by either ClaD or CitB (Figure 14a). The only substrate that showed no activity with both ClaD and CitB was lacking any substituents at R¹, indicating that an electron-withdrawing group is essential at this position. These benzylic alcohols could be further functionalized by gentle heating and addition of various alcohols, amines, thiols, and alkenes to the reaction mixture. This process is compatible with chemoselectively labeling cysteine-containing peptides in situ, which could be applicable for biorthogonal chemistry applications. These reactions could be scaled up to >500 mg with no enzyme purification required. This mild one-pot process was also applied to chemoenzymatically synthesize (–)-xyloketal D (4) and its diastereomer (103) from 101 in a 2:1 ratio with overall improved yields compared to any previously reported total synthesis (Figure 14b). This strategy is amenable to chemoenzymatically synthesize chromane-containing natural products and derivatives alike.

Figure 14.

Chemoenzymatic benzylic functionalization reactions via an o-QM with α-KG-dependent nonheme iron oxygenases ClaD and CitB. (a) Various aryl-substituted substrates could be accepted. Additional nucleophiles could be added to the reaction mixture after benzyl alcohol protection. (b) One-pot chemoenzymatic synthesis of (–)-xyloketal D (4).

CONCLUSIONS AND FUTURE DIRECTIONS

ortho-Quinone methides are notoriously reactive chemical intermediates that require careful manipulation to achieve desired chemo-, regio-, and stereoselective outcomes. Numerous chemical approaches have been designed to harness o-QMs in natural product total syntheses but face many obstacles to be effectively utilized. Nature has refined its own strategies to generate o-QMs with greater chemo-, regio-, and stereoselectivity. Recent advancements in sequencing technology and synthetic biology have revealed 11 enzymes across four distinct families thus far that can generate o-QMs and perform oxidative cyclizations, intra- and intermolecular nucleophilic addition reactions, dehydrogenations, and hetero-Diels–Alder reactions. The remarkable enzyme and substrate diversity reported so far is a promising starting point for engineering enzymes for biocatalytic o-QM generation and utilization. Furthermore, the presently known enzymes can be used as probes in genome mining efforts to identify similar enzymes with complementary reactivity, improved catalytic activity, and to aid in the discovery of new biosynthetic gene clusters.