Exploiting the Potential of Meroterpenoid Cyclases to Expand the Chemical Space of Fungal Meroterpenoids

Abstract

Fungal meroterpenoids are a diverse group of hybrid natural products with impressive structural complexity and high potential as drug candidates. In this work, we evaluate the promiscuity of the early structure diversity-generating step in fungal meroterpenoid biosynthetic pathways: the multibond-forming polyene cyclizations catalyzed by the yet poorly understood family of fungal meroterpenoid cyclases. In total, 12 unnatural meroterpenoids were accessed chemoenzymatically using synthetic substrates. Their complex structures were determined by 2D NMR studies as well as crystalline-sponge-based X-ray diffraction analyses. The results obtained revealed a high degree of enzyme promiscuity and experimental results which together with quantum chemical calculations provided a deeper insight into the catalytic activity of this new family of non-canonical, terpene cyclases. The knowledge obtained paves the way to design and engineer artificial pathways towards second generation meroterpenoids with valuable bioactivities based on combinatorial biosynthetic strategies.

Introduction

Fungal meroterpenoids have earned significant interest from the scientific community as well as from the pharmaceutical and chemical industry due to their remarkable structural architectures and potent bioactivities.^[1] Pyripyropene A (1), isolated from Aspergillus fumigatus, is the strongest known inhibitor of acyl-CoA:cholesterol acyltransferase and is being developed for the treatment of atherosclerosis.^[2] Additionally, 1 exhibits insecticidal properties and a commercial insecticide based on the pyripyropene core structure has been recently marketed.^[3] Derivatives of mycophenolic acid (2), isolated from Penicillium sp., are clinically used immunosuppressant drugs and inhibit inosine 5′-monophosphate dehydrogenase.^[4] Andrastin A (3) from Penicillium albocoremium is an inhibitor of protein farnesyl transferase and a potent anti-cancer agent,^[5] whereas tropolactone D (4) from Aspergillus sp. is a cytotoxic agent against human colon carcinoma (Figure 1A).^[6] The genetic basis for fungal meroterpenoid biosynthesis has only been elucidated in recent years, with the first biosynthetic gene cluster reported for pyripyropene A in 2010.^[7]

Figure 1.

A) Selected examples of fungal meroterpenoids. B) Proposed catalytic mechanism of meroterpenoid cyclases exemplified by Pyr4-mediated reaction of epoxyfarnesyl-HPPO (5) to deacetyl pyripyropene E (6). The polyketide portion is shown in green, terpenoid in blue.

Since then, the discovery of several related gene clusters revealed a common modular assembly logic for all meroterpenoid pathways, composed of i) building block generation (polyketide synthase, oligoprenyl synthase), ii) assembly of building blocks (prenyltransferase), iii) early structural diversification by an epoxidase enzyme followed by action of a novel family of terpene cyclases and iv) late stage diversification by tailoring enzymes such as cytochrome P450 monooxygenases and α-ketoglutarate-dependent dioxygenases.^[8] Their strong biological activities as well as the conserved modular logic of their biosynthetic pathways make meroterpenoids attractive targets for artificial pathway engineering to generate novel structures with new and improved activities. Herein, we set out to evaluate the potential of meroterpenoid cyclases to generate novel scaffolds by employing natural and unnatural synthetic substrate analogues. The results obtained provide valuable information on matching pathway combinations regarding the interchange-ability of employed meroterpenoid cyclases. At the same time, chemoenzymatic access to eight new scaffolds, thus far unprecedented from natural sources or chemical, biomimetic polyene cyclizations, could be achieved.

Results and Discussion

Targeted Enzymes and Synthesis of Substrates

Non-canonical terpene cyclases involved in meroterpenoid biosynthesis are integral membrane proteins of compact size (ca. 25 kDa).^[7,8] Mutagenesis studies on the model cyclase Pyr4 involved in the pyripyropene biogenesis revealed two highly conserved acidic amino acid residues (Glu63 and Asp218) crucial for enzyme function which are proposed to initiate polyene cyclization by protonation of the terminal epoxide function of the substrate thereby triggering subsequent polyene cyclization (Figure 1B). The mechanism resembles that of type-II terpene synthases of the 2,3-oxidosqualene-lanosterol cyclase type; however, protein structural data to evaluate the mechanism of meroterpenoid cyclases is still lacking.^[7–9] Phylogenetic analysis of characterized meroterpenoid cyclases shows a close relation to the group of Pyr4-like synthases involved in fungal indole diterpene biosynthesis (LtmB, AtmB, and PaxB) and a distant relation to the bacterial enzymes XiaE and DmtA1 (Figure 2).^[10] The protein sequence alignment for selected enzymes is shown in Figure S6A.

Figure 2.

Phylogenetic analysis of reported meroterpenoid cyclases and their respective native substrates.

The substrates of known meroterpenoid cyclases are composed of a linear epoxyoligoprenyl chain, in most cases derived from farnesyl diphosphate (FPP) and a distinct non-terpenoid portion usually generated by a designated polyketide synthase (Figure 2). These biosynthetic intermediates are difficult to obtain from enzymatic reactions, since typically low concentrations are observed and the frequently employed heterologous expression host Aspergillus oryzae contains endogenous hydrolases producing high amounts of a shunt diol product.^[7] To overcome this limitation, we recently developed a modular synthesis of the widespread 3,5-dimethylorsellinic acid (DMOA)-containing substrate family. The methodology involves base-mediated, regioselective dearomatization of DMOA with farnesyl electrophiles.^[11] We expanded the synthetic scope by employing enantiopure (10R)- and (10S)-epoxyfarnesyl building blocks in this reaction to obtain the naturally occurring substrates (10R)- and (10S)-(2E,6E)-5′-DMOA methyl ester (7a, 7b), as well as (10R)- and (10S)-(2E,6E)-3′-DMOA methyl ester (8a, 8b). Additionally, we also accessed unnatural substrates including the (2Z,6E)-epoxyfarnesyl congeners (10R)- and (10S)-(2Z,6E)-5′-DMOA methyl ester (9a, 9b) and (10R)- and (10S)-(2Z,6E)-3′-DMOA methyl ester (10a, 10b) using the same strategy (Figure S1, Figure 3). As dearomative alkylation of DMOA leads to formation of two inseparable diastereoisomers with respect to 3′ and 5′ dearomatization, the substrates obtained were used as diastereomeric mixtures, with fixed stereochemistry for the epoxide moiety (88–98% ee). With these substrates in hand, we targeted nine reported meroterpenoid cyclases: Pyr4,^[7] CdmG,^[12] AndB,^[13] AdrI′,^[14] NvfL,^[15] PrhH,^[16] Trt1,^[17] AscF,^[18] and MacJ.^[19]

Figure 3.

Synthetic approach and structures of obtained and tested DMOA substrates (7a/7b–10a/10b).

As we did not gain access to the MacJ producer strain, we cloned the homologous protein MacJ′ from Penicillium chrysogenum MT-12 (96% identity). The intron-free genes were expressed in the heterologous host Saccharomyces cerevisiae INVSc1 and cell free extracts were prepared and utilized for in vitro reactions with synthetic substrates 7a/7b–10a/10b. Monitoring of the reactions by HPLC revealed a surprisingly high degree of promiscuity as several new products were detected (Figure S2–S5, Table 1).

Table 1:

Overview on productive enzyme substrate combinations.

	(2E,6E)-5′-DMOA	(2E,6E)-3′-DMOA	(2Z,6E)-5′-DMOA	(2Z,6E)-3′-DMOA
Pyr4	(10S)-7a	(10S)-8a	–	(10S)-10a
MacJ′	(10S)-7a	(10S)-8a	–	(10S)-10a
CdmG	(10S)-7a	(10S)-8a	(10S)-9a	(10S)-10a
AndB	(10S)-7a	(10S)-8a	(10S)-9a	–
Trt1	(10S)-7a (10R)-7b[a]	–	(10R)-9b	–
AdrI′	(10S)-7a (10R)-7b[a]	–	(10R)-9b	–
PrhH	(10R)-7b[a]	–	–	–
NvfL	–	–	–	–
AscF	–	–	–	(10S)-10a

^[a]Native enzyme substrate combination.

Substrate Scope of Pyr4 and MacJ′

Pyr4, a cyclase which naturally utilizes the (10S)-configured epoxide 5 (Figure 1B), was found to accept the substrates (10S)-(2E,6E)-5′-DMOA (7a), (10S)-(2E,6E)-3′-DMOA (8a) as well as (10S)-(2Z,6E)-3′-DMOA (10a) based on the detection of newly formed peaks in the HPLC chromatogram (Table 1, Figure S2–S5). To elucidate the structures of the putative new enzyme products, we carried out preparative scale transformations in which case reaction of Pyr4 with substrate 7a led to the isolation of compounds 11 and 12 (Scheme 1A).

Scheme 1.

Structures of isolated meroterpenoids obtained from Pyr4- and MacJ′-mediated reactions with A) (10S)-7a; B) (10S)-8a; C) (10S,3′R)-10a.

Both products exhibit a chair-chair conformation for the A/B ring system, as is also found for the natural cyclization to form pyripyropene E (6). However, the terminating cation-quenching step differs for both products, leading to 11 after C–O bond formation and 12 after C–C bond formation. These products were also recently identified from chemical cyclization of rac-7 by using EtAlCl₂/Et₂AlCl as Lewis acid promoter.^[11] The [3.3.1] bridged structure in 12 is also found in asperterpenes A and B, recently isolated and potent BACE1 inhibitors from Aspergillus terreus.^[20] As can be delineated from the configuration of position 5′, Pyr4 is able to accept both diastereomers, (10S,5′S)-7a and (10S,5′R)-7a, to form 11 and 12, respectively. The findings suggest that Pyr4, which usually accepts the bulkier substrate 5, exhibits some degree of promiscuity towards changes in the polyketide portion. This is further demonstrated by the successful conversion of (10S,3′R)-8a to 13, as well as (10S,3′S)-8a to 14. Substrate 8a carries the epoxyfarnesyl chain connected to the 3′-position of the DMOA-building block instead of the 5′-position as found in 7a. Products 13 and 14 are also accessed from chair-chair substrate conformations, as observed for the natural reaction of Pyr4 towards the chair-chair product 6. These findings indicate that the enzyme active site cavity provides a template for precise arrangement of substrate conformation and guidance of a stereochemically distinct polyene cyclization (Scheme 1B). Their structures were determined by 1D and 2D NMR analyses, and further confirmed by the recently developed crystalline sponge (CS) method which enables “crystallization-free” X-ray crystallography. The method makes use of crystalline porous metal complexes that can absorb and orient a compound of interest. The neat alignment of compounds in the pores of the complex makes them observable by conventional X-ray structure analysis without the need for prior crystallization.^[21]

We were also interested in the flexibility of Pyr4 towards changes in the farnesyl chain and therefore subjected substrate analogues 9a/9b and 10a/10b to Pyr4. Indeed, Pyr4 was able to convert (10S,3′R)-10a to the new meroterpenoid 15 bearing a cis-fused B/C ring system, presumably derived from a chair-chair substrate conformation (Scheme 1C). The natural substrate of Pyr4 contains an (S)-configured epoxide functionality. Interestingly, all productive enzyme substrate combinations were restricted to the (10S)-series of substrates, as none of the analogous (10R)-epoxy substrates (10R)-7b, (10R)-8b, (10R)-9b or (10R)-10b were accepted. This finding indicates a strict recognition of the epoxide within the substrate binding site of the enzyme. In addition, we found that MacJ′ can also produce products 13 and 14 from (10S)-8a and 15 from (10S,3′R)-10a (Scheme 1B/C). MacJ, naturally involved in the biosynthesis of the drimane meroterpenoid macrophorin A, is one of two known Pyr4-like cyclases which do not require substrate activation by epoxidation. Instead, MacJ is able to directly protonate the terminal double bond to initiate cyclization (Figure 2, Figure S6B). It is therefore interesting to note that MacJ′ also exhibits a clear preference for the (10S)-stereoisomers of epoxide substrates evaluated (Table 1, Figure S2–5).

Substrate Scope of CdmG and AscF

A meroterpenoid cyclase which is phylogenetically closely related to Pyr4 is CdmG (Figure 2), utilized in the biosynthetic pathway towards chrodrimanins from Penicillium verruculosum.^[12] Chrodrimanins exhibit strong inhibitory activities against protein tyrosine phosphatase 1B (PTP1B) and are potential drug candidates for the treatment of type 2 diabetes and obesity.^[22] CdmG naturally catalyzes the formation of 3-hydroxypentacecilide A (17) from (S)-configured epoxide 16. In contrast to Pyr4, CdmG controls the substrate conformation in a chair-boat manner, leading to an inverted stereochemical outcome for the trans-configured B/C ring system (Scheme 2A). With this apparent change in conformational control, we were interested in determining the substrate promiscuity of CdmG and the putative differences in product formation.

Scheme 2.

A) Natural Reaction of CdmG and isolated meroterpenoids obtained from CdmG-mediated reactions with B) (10S,3′S)-8a; C) (10S,5′S)-9a; D) (10S)-10a.

When CdmG was incubated with synthetic substrates, a high degree of promiscuity was also observed as 7a, 8a, 9a, and 10a were accepted by CdmG and led to the production of new products (Figure S2–S5, Table 1). Whereas products from substrate (10S)-7a were found to be too unstable for structural characterization, products from substrates 8a, 9a, and 10a were successfully isolated and structurally characterized. Reaction with (2E,6E)-configured substrate (10S)-8a led to formation of compound 18 derived from the (10S,3′S)-8a isomer. The structure of 18 was determined by 2D NMR analysis and was further supported by X-ray-CS-diffraction data. The stereochemical outcome of the cyclization indeed demonstrated a conserved chair-boat conformational control of the substrate by the enzyme (Scheme 2B).

Reaction of CdmG with (10S)-9a represents the only case where the two products found were derived from one substrate stereoisomer, in this case (10S,5′S)-9a, leading to the isolation of 19 and 20. Whereas 19 is produced from a C–O bond forming event, product 20 is derived from C–C bond formation.

In both cases, the same trans-cis configuration is observed for the A/B/C-ring system (Scheme 2C). Compounds 19 and 20 have also recently been identified from chemical cyclization where it was further shown that 20 can be rearranged to 19 by formic acid treatment.^[11] To confirm the enzymatic origin of 19, 20 was incubated under enzyme reaction conditions (KPP pH 7.4, 16 h, 30°C) but was found to not interconvert to 19. Reaction of CdmG with (2Z,6E)-configured substrate 10a produced the new meroterpenoids 21 and 15, derived from a chair-boat substrate conformation (major) and a chair-chair substrate conformation (minor), respectively. As observed for the promiscuity of Pyr4, CdmG also had a strict preference for the epoxide stereoconfiguration, as none of the (10R)-epoxides were accepted by CdmG.

AscF from the ascochlorin pathway (Figure S6B) was found to also produce 21 and small amounts of 15 from (10S)-10a as the only accepted substrate (Table 1, Figure S2–S5) and thus represents a meroterpenoid cyclase with very low tolerance towards alternative substrates.

Substrate Scope of AndB

The meroterpenoid cyclase AndB from the anditomin pathway^[13] utilizes DMOA-derived substrate 22 with an (S)-configured epoxide to produce preandiloid A (23) with a chair-boat conformation (Scheme 3). In contrast to the results discussed for CdmG, which also controls the conformation in a chair-boat manner, AndB was found to exhibit a differing selectivity based on the 3′- and 5′-stereocenters and therefore was found to produce different products (Table 1, Figure S2–S5). Specifically, AndB was found to accept (10S,3′R)-8a to produce the novel meroterpenoid 24 (Scheme 3B). Similar to CdmG and Pyr4, the native conformational control of the prenyl chain was conserved, as a chair-boat substrate arrangement was evident leading to 24. AndB also accepted the (2Z,6E)-configured substrate (10S,5′R)-9a leading to the isolation of meroterpenoid 25. The structure elucidation for 25 was challenging as 2D NMR analysis did not clearly reveal the connectivity between the terpenoid and non-terpenoid portions. Additionally, the relative configuration between the A/B and C/D ring systems were difficult to determine due to ambiguous NOESY correlations. However, we were able to fully establish the structure of 25 using crystalline sponge-X-ray analysis which revealed the presence of an unprecedented 6-5-ring system connected to a 5–6 ring system via a single C–C bond (Scheme 3C).

Scheme 3.

A) Natural reaction of AndB and structures of isolated meroterpenoids obtained from AndB-mediated reactions with B) (10S,3′R)-8a; C) (10S,5′R)-9a.

DFT Calculations for the Formation of 25, 20, CC, and 19

The latter finding was surprising, as all reactions in this study lead to the formation of 6-6-ring systems for the A/B rings and also no natural cyclization towards 6-5-ring systems has been reported thus far. To gain further insight into the cyclization mechanism for the formation of 25 from (10S,5′R)-9a by AndB, we conducted computational studies using density functional theory (DFT) at the B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) and CPCM(H₂O)-B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) levels^[23] (see SI for details). The results (Figure 4) revealed that the first intermediate in the reaction cascade (modelled here in the absence of enzyme) is the monocyclic tertiary cation A, generated by an endergonic process via transition state 9a-TS. A is then converted, 010404e, to 25-H⁺ by an exergonic concerted process consisting of formation of the 5-membered B-ring in tandem with nucleophilic attack of the adjacent oxygen functionality via transition state A-TS. Inclusion of implicit solvent (CPCM(H₂O), in parentheses; single point calculations on previously optimized geometries in gas phase) led to lower barriers (Figure 4).^[24,25] With either a nonpolar (gas phase) or polar (water) surroundings, the barrier for initial cyclization is high for a biological reaction,^[26] however, in the absence of enzyme, reactant (10S,5′R)-9a-H⁺ relaxes to a non-productive conformation with an internal hydrogen-bond between the alcohol and epoxide; consequently, conformational preorganization by the enzyme should lower the barrier and this could be assisted by specific oriented noncovalent interactions with the transition state structure. An alternative mechanism for formation of the second ring could involve Markovnikov addition to form a 6–6 intermediate, followed by ring-contraction in concert with tetrahydro-furan ring formation. We find, however, that Markovnikov addition leads directly to a “6-6-6-6” product (C) that is not experimentally observed. We were able to optimize tertiary carbocation B as a minimum and this species can then undergo ring contraction to yield 25-H⁺, but accessing B would require escape from the deep energy well associated with C and a large conformational change. The enzyme would, however, have to distinguish between A-TS and A′-TS, again by conformational biasing and/or well-placed noncovalent interactions with the transition state structure.

Figure 4.

DFT calculations for the cyclization mechanism of (10S,5′R)-9a towards 25 by AndB. Two conformers of A were located, one connected to 9a-TS and another connected to A-TS, which differ in energy by 0–3 kcalmol⁻¹, depending on the level of theory; see SI for details. Computed (B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) in black (top), CPCM(H₂O)-B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) in parentheses (bottom)) relative free energies (kcalmol⁻¹, italics) for minima and transition state structures (TSSs) involved in formation of compound 25. Bond distances are in Angstroms [Å].

Mechanisms for formation of 20 (Figure 5) and CC (Figure 6) from protonated stereoisomers of 9a were also subjected to computational analysis. For both reactions, we found highly asynchronous but concerted pathways in which no discrete carbocationic intermediates are formed. Similar concerted polycyclizations have been reported for related systems.^[27] Both reactions also are predicted to be essentially barrierless once productive reactant conformations are attained, suggesting that preorganization controls which product is formed by a given enzyme. Conversion of 20-H⁺ to 19-H⁺ is predicted to be an endergonic process (Figure 7; neutral 19 is predicted to be several kcal mol⁻¹ lower in energy than neutral 20, however^[11]), but an appropriately positioned base in a restricted enzyme active site could selectively deprotonate 19-H⁺. The barrier for the 20-H⁺ to 19-H⁺ interconversion is also less certain than others described above, since this reaction involves asynchronous bond-breaking, C–C bond rotation, and bond-making that leads to a “loose” transition state for which entropy and the effects of externally-imposed conformational constraints are difficult to estimate.

Figure 5.

DFT calculations for the cyclization mechanism for the formation of 20. Computed (B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) (on top), CPCM(H2O)-B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) (in parentheses) and MPW1PW91/6-31G(d,p)//B3LYP/6-31G(d,p) [in brackets]) relative free energies (kcalmol⁻¹, italics) for minima and transition state structures (TSSs) and IRC traces for the respective TSSs. Bond distances are in Angstroms [Å].

Figure 6.

DFT calculations for the cyclization mechanism for the formation of CC. Computed (B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) (on top), CPCM(H2O)-B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) (in parentheses) and MPW1PW91/6-31G(d,p)//B3LYP/6-31G(d,p) [in brackets]) relative free energies (kcalmol⁻¹, italics) for minima and transition state structures (TSSs) and IRC traces for the respective TSSs. Bond distances are in Angstroms [Å].

Figure 7.

DFT calculations for the interconversion of 20-H⁺ to 19-H⁺. Computed (B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) (on top), CPCM(H₂O)-B3LYP-D3(BJ)/6-31G(d,p)//B3LYP/6-31G(d,p) (in parentheses) and MPW1PW91/6-31G(d,p)//B3LYP/6-31G(d,p) [in brackets]) relative free energies (kcalmol⁻¹, italics) for minima and transition state structures (TSSs). Bond distances are in Angstroms [Å].

Substrate Scope of Trt1, AdrI′, PrhH, and NvfL

Another phylogenetic clade is formed by Trt1, AdrI, AdrI′, AusL, and PrhH (Figure 2). These enzymes share the DMOA substrate (10R,5′R)-7b, but differ with regard to their product specificity. Trt1 catalyzes the formation of preterretonin (27) via a chair-chair-chair substrate conformation forming intermediary cation 26, followed by Wagner–Meer-wein rearrangement and a terminating deprotonation of H_a. AdrI shares the intermediary cation 26 and rearrangement but differs in the terminating deprotonation side (H_b) producing andrastin E (28). AusL and PrhH both catalyze the formation of protoaustinoid A (29) from 26 after direct deprotonation of H_c (Scheme 4A). NvfL from the novofumigatonin pathway utilizes a highly similar substrate as Trt1, AdrI, AusL, and PrhH, carrying a free carboxylic acid instead of the methyl ester in the DMOA-unit. The free acid is crucial for enzyme function and the protein catalyzes formation of a spiro-center (Figure S6B).

Scheme 4.

A) Natural reaction of Trt1, AdrI/AdrI′ and AusL/PrhH. B) Isolated meroterpenoids obtained from Trt1-mediated reactions with (10S,5′R)-7a.

Consistent with the obvious tight recognition of the polyketide portion necessary to achieve these sophisticated and distinct reactions, a comparable low promiscuity for this group of integral membrane-bound enzymes was observed. Whereas PrhH and NvfL did not accept any of the tested unnatural substrates, both Trt1 and AdrI′ accepted (2Z,6E)-9b, with the natural 5′-substitution pattern in the polyketide portion (Table 1, Figure S2–S5).

However, for these cases products were found to be too unstable for structure determination. To our surprise, we found that Trt1 as well as AdrI′ were able to convert (10S)-7a, the native substrate with inverted stereochemistry with respect to the epoxide functionality. We succeeded in the isolation of the Trt1-mediated product from (10S,5′R)-7a and the structure was determined to be 3-epi-preterretonin (30). The formation of this product can be envisioned to occur via a boat-chair substrate conformation (Scheme 4B).^[28]

Enzyme Kinetics of Trt1

To shed additional light on the substrate promiscuity of meroterpenoid cyclases, we were interested in the comparison of the K_M values of the natural substrate to unnatural substrate analogous. Synthetic (10R,5′R)-7b is the natural substrate of Trt1, which was also found to accept unnatural (10S,5′R)-7a, and (10R)-9b. Since Trt1 and other meroterpenoid cyclases cannot be purified, the integral membrane bound enzyme was used as a crude enzyme preparation. To ensure comparable reaction conditions, Trt1 was freshly prepared and used for all kinetic assays on the same day. The results revealed an apparent K_M value of 34 μM for the natural substrate 7b, whereas a 3-fold (95 μM) and 5-fold (144 μM) higher value was found for substrates 7a and 9b, respectively. The V_max values were determined as 22 μMmin⁻¹, 7 μMmin⁻¹ and 0.5 μMmin⁻¹, for 7b, 7a and 9b, respectively. These findings indicate that the unnatural substrates have a lower affinity for the enzyme, but the values are of the same order of magnitude and thus further demonstrate the promiscuity observed for this enzyme class.

Conclusion

In summary, we have demonstrated chemoenzymatic access to twelve complex, unnatural DMOA-derived meroterpenoids of which eight represent novel compounds by exploiting the surprisingly high promiscuity of fungal meroterpenoid cyclases.

Synthetic 3,5-dimethylorsellinic acid (DMOA)-containing substrates were prepared using dearomative alkylation and evaluated against nine meroterpenoid cyclases. The results demonstrate tight recognition of the epoxide functionality by the cyclase panel, but tolerance towards the polyketide portion. The conserved cyclization mechanism initiated by epoxide protonation and rigid control of the substrate conformation in the cyclase active site cavity led to the formation of several variations of naturally occurring meroterpenoid scaffolds, as well as generation of a completely new scaffold, although the number of cyclization steps that are concatenated into concerted processes appears to be system-dependent. The challenging structure elucidation and determination of relative and absolute configurations for the obtained enzyme products was solved by combining 2D NMR data analysis with the recently developed method of crystal-line-sponge-X-ray diffraction analysis. This method allowed us to access the crystallographic data of non-crystalline enzyme products such as 25 for unambiguous structure determination and thus represents a powerful technique in combination with second generation natural product discovery.

The knowledge obtained in this study can be used for the design of artificial pathways by recombining biosynthetic genes in a suitable heterologous production host. Whereas productive enzyme combinations with naturally occurring substrates such as 7a and 7b provide a straightforward access by reconstituting known pathways towards these substrates and interchanging the introduced meroterpenoid cyclase, access to (2Z,6E)-configured prenyl substrates require further engineering efforts, for example, by employing a known bacterial (2Z,6E)-selective FPP synthase from Mycobacterium^[29] and optimization of downstream enzymes towards substrates like 9a and 9b. Furthermore, the new meroterpenoids obtained can be evaluated as substrates for downstream tailoring enzymes such as α-ketoglutarate-dependent dioxygenases. Since several protein crystal structures have been reported in recent years, protein engineering by either rational strategies or directed evolution represents an exciting opportunity to create novel “unnatural” meroterpenoids with valuable biological properties.^[30]