Using Terpene Synthase Plasticity in Catalysis: On the Enzymatic Conversion of Synthetic Farnesyl Diphosphate Analogues.

Four synthetic farnesyl diphosphate analogues were enzymatically converted with three bacterial sesquiterpene synthases, including β-himachalene synthase (HcS) and (Z)-γ-bisabolene synthase (BbS) from Cryptosporangium arvum, and germacrene A synthase (SmTS6) from Streptomyces mobaraensis. These enzyme reactions not only yielded several previously unknown compounds, showing that this approach opened the door to a new chemical space, but substrates with blocked or altered reactivities also gave interesting insights into the cyclisation mechanisms and the potential to catalyse reactions with different initial cyclisation modes.

Introduction

Terpenoids are the largest class of natural products and are widely distributed in all kingdoms of life where they fulfill diverse biological functions. Their structural diversity is controlled by terpene synthases (TPSs), which catalyse the conversion of acyclic oligoprenyl diphosphates into structurally complex terpene hydrocarbons or alcohols.[ , ] Monoterpene synthases (MTPSs) can convert geranyl diphosphate (C10, GPP) into monoterpenes, sesquiterpene synthases (STPSs) catalyse the transformation of farnesyl diphosphate (C15, FPP) into sesquiterpenes, and diterpenes synthases (DTPSs) and sesterterpene synthases (StTPSs) can use geranylgeranyl diphosphate (C20, GGPP) and geranylfarnesyl diphosphate (C25, GFPP) to form diterpenes and sesterterpenes, respectively.[ , ]

Besides these regular cases, some non‐canonical TPSs naturally accept modified substrates, e. g. the biosynthesis of 2‐methylisoborneol (1) proceeds through methylation of GPP to 2‐Me‐GPP, followed by cyclisation by the 2‐methylisoborneol synthase (MIBS, Scheme 1).[ , , ] Other recently described examples include the biosynthesis of sodorifen (3) that starts with a methylation induced cyclisation of FPP to presodorifen diphosphate (2) by SodC followed by conversion into 3 by SodD, or the biosynthesis of longestin involving methylation of IPP to (Z)‐4‐methyl‐IPP by Lon23 and specific incorporation into (4R,12R)‐4,12‐dimethyl‐GGPP (dmGGPP) by Lon22. Furthermore, a santalene and bergamotene synthase (SBS) from the wild tomato Solanum habrochaites has been reported that naturally converts (2Z,6Z)‐FPP into a mixture of sesquiterpenes including (+)‐α‐santalene (4), (+)‐endo‐β‐bergamotene (5), and (−)‐endo‐α‐bergamotene (6).

Scheme 1

Non‐canonical terpenoid biosynthesis.

These natural systems raise the question whether also canonical TPSs have the potential to convert substrates other than the regular oligoprenyl diphosphates. Recent research has demonstrated that this is indeed the case, revealing that e. g. halogenated substrates,[ , ] substrates with additional or missing Me groups or altered methylation pattern,[ , , , ] functional groups attached to[ , ] or heteroatoms inserted into the chain, with hydrogenated double bonds,[ , ] or stereoisomers with Z‐configured olefins such as (2Z,6E)‐FPP can be converted.[ , ] The substrate modifications can block reactivity, as for compounds with hydrogenated olefins, or open new reaction pathways as for compounds with altered methylation patterns in which cations can be stabilised at different carbons of the isoprenoid chain. Also functional groups may directly engage in the TPS catalysed reaction. Here we report on the synthesis of four new FPP analogues and their enzymatic conversion with β‐himachalene synthase (HcS) and (Z)‐γ‐bisabolene synthase (BbS) from Cryptosporangium arvum, and germacrene A synthase (SmTS6) from Streptomyces mobaraensis. For HcS cyclisation cascades from FPP to β‐himachalene (7) with initial 1,11‐ (blue path in Scheme 2A) or 1,6‐cyclisation (red path) can be formulated. The co‐occurrence of side products arising by 1,11‐cyclisation (9‐epi‐β‐caryophyllene and γ‐humulene) with simultaneous absence of any 1,6‐cyclised products may favour the pathway through initial 1,11‐cyclisation for 7. BbS catalyses a 1,6‐cyclisation of FPP via nerolidyl diphosphate (NPP) and the bisabolyl cation (A 3) into (Z)‐γ‐bisabolene (8, Scheme 2B). Previous incubation experiments with (R)‐ and (S)‐NPP have demonstrated that this process involves the intermediates (R)‐NPP and (S)‐A 3. SmTS6 converts FPP through 1,10‐cyclisation to the (E,E)‐germacradienyl cation A 4 into germacrene A (9, Scheme 2C).

Scheme 2

Cyclisation reactions with FPP. Cyclisation to A) β‐himachalene (7) by HcS, B) (Z)‐γ‐bisabolene (8) by BbS, and C) germacrene A (9) by SmTS6.

Results and Discussion

For investigations with the selected enzymes the FPP analogues 10–13 were designed. Substrate 10 has a saturated bond instead of the terminal double bond of FPP, in analogue 11 the position of the terminal double bond is shifted, and in substrates 12 and 13 a methyl group at the middle or the terminal double bond of FPP is removed and exchanged by a ketone group (Scheme 3A). We hypothesised that substrate 10 may be converted smoothly with BbS, but cannot react in a 1,10‐cyclisation and may thus only give acyclic products with SmTS6. With HcS further insights into the question of initial 1,6‐ versus 1,11‐cyclisation may be obtained. Analogues 11 and 12 could undergo the usual 1,6‐cyclisation with BbS, but with SmTS6 and HcS new reaction paths may be opened. Finally, with 13 new reaction paths may be observed with BbS, while with SmTS6 a 1,10‐cyclisation could still be possible, and for HcS the result may depend on the question of initial 1,6‐ or 1,11‐cyclisation, potentially leading to new reaction paths. The FPP analogues 10–12 were synthesised as reported previously,[ , ] and 13 was synthesised through Corey‐Seebach umpolung from 1,3‐dithiane 14 and iodide 15 to yield 16 (Scheme 3B). Deprotection via 17 to 18 followed by bromination and phosphorylation yielded 13.

Scheme 3

A) FPP analogues 10–13 used in this study. B) Synthesis of FPP analogue 13, for reaction conditions cf. Scheme S1.

In a first series of experiments, 10–13 were enzymatically converted with HcS (Figures S1 and S2). This enzyme did not accept 10 as substrate, suggesting that HcS does not catalyse an initial 1,6‐cyclisation, but proceeds through 1,11‐cyclisation, in agreement with our previous report. With substrate 11 one major product 19 was obtained (Scheme 4A) that was isolated and characterised as (1Z,5E,9E)‐1,5,9‐trimethylcyclododeca‐1,5,9‐triene (19) by NMR spectroscopy (Table S1, Figures S3–S10). The 1H NMR spectrum of 19 showed line broadening for all CH2 groups, pointing to slowly interconverting conformers, but sharp signals in the 13C NMR spectrum. As a result of line broadening, signals for the hydrogens attached to C12 were missing in all 1H based spectra. Therefore, to secure the structure of 19 a catalytic hydrogenation was performed that yielded an inseparable mixture of the two possible diastereomers, C 3v symmetric 20 and C s symmetric 21 (Scheme 4B), that were observed by 13C NMR and GC/MS (Figures S11 and S12). Compound 19 can be formed from 11 through a newly opened reaction path that is not possible for FPP, i. e. by 1,12‐cyclisation, leading to a cationic intermediate with a tertiary cation at C11 in A5, followed by deprotonation to 19 (Scheme 4A).

Scheme 4

A) Enzymatic conversion of FPP analogue 11 with HcS. B) Catalytic hydrogenation of 19.

Substrate 12 was converted by HcS into multiple products. The main compound was isolated and identified as 22 ([α]D 25=−16.2, c 0.21, CH2Cl2), the enantiomer of ent‐22 ([α]D 20=+51.7, c 0.12, CH2Cl2) that we had previously obtained from 12 with dauc‐8‐en‐11‐ol synthase (DcS) from Streptomyces venezuelae (Scheme 5). The formation of 22 requires isomerisation to (R)‐A6 and subsequent 1,6‐cyclisation by anti‐SN2’ reaction, resulting in (S)‐B 6. Final attack of water yields 22. These results show that HcS can also catalyse a 1,6‐cyclisation with substrate 12.

Scheme 5

Enzymatic conversion of FPP analogue 12 with HcS.

FPP analogue 13 yielded with HcS compounds 23 and 24 (Scheme 6A). Both compounds were isolated and their structures elucidated by NMR spectroscopy (Tables S2 and S3, Figures S13–S28). The absolute configuration of 23 was determined by chemical correlation with both enantiomers of nerolidol (25, Scheme 6B). Therefore, (S)‐ and (R)‐25 were converted by catalytic hydrogenation, yielding silica gel chromatographically inseparable diastereoisomeric mixtures of (3R,7RS)‐26 a and (3S,7RS)‐26 b that were also inseparable by GC on a chiral stationary phase (Figure S29). However, the retention times for 26 a and 26 b were clearly different, allowing to conclude on the configuration at C3. Compound 23 was converted by Wittig reaction into 27 (Table S4, Figures S30–S37) followed by catalytic hydrogenation that yielded a sample identical to 26 b with minor formation of 26 a (3 %). Thus, the enzyme product is (R)‐23 (94 % ee, Figure S29). Its formation can be explained from 13 either directly or by isomerisation to (S)‐A 7 and hydrolysis with inversion of configuration at C3.

Scheme 6

A) Enzymatic conversion of FPP analogue 13 with HcS. B) Correlation of (R)‐23 with both enantiomers of nerolidol (25). C) Degradation of 24 by ozonolysis.

The formation of 24 can be explained by 1,11‐cyclisation of A 7 to B 7 with anti‐SN2’ attack at C1. A reprotonation induced cyclisation of C 7 to D 7 and deprotonation yields 24. If (S)‐A 7 is a common intermediate for both products 23 and 24, the shown absolute configuration of 24 may result (Scheme 6A). Its ozonolysis to 28 (Table S5, Figures S38–S45) was followed by trials to convert this compound into the bis‐dinitrophenylhydrazone for crystallisations. Unfortunately, this approach was not successful and clarification of the absolute configuration of 28 is open.

In a second series of experiments, the substrate analogues 10–13 were enzymatically converted with BbS (Scheme 7, Figures S46–S47). The FPP analogues 10–12 can be converted through 1,6‐cyclisation into compounds 29–31 (Tables S6–S8, Figures S48–S71), while 13 yielded the acyclic product 23. These reactions proceed with similar efficiency as observed for the native substrate FPP, showing that the structural modifications did not influence acceptance by BbS. Based on the cyclisation mechanism of BbS with FPP, (R)‐A 7 in analogy to (R)‐NPP may be the intermediate which should further react by abstraction of diphosphate and SN2 attack of water with inversion of configuration to yield (S)‐23. The absolute configuration of 23 was confirmed by analysis on a chiral GC column, compared to (R)‐23 obtained from 13 with HcS (Figure S72).

Scheme 7

Enzymatic conversions FPP analogues 10–13 with BbS.

The FPP analogues 10–13 were finally tested with SmTS6 (Figures S73 and S74). With substrate 10, only two acyclic products (6E)‐10,11‐dihydro‐β‐farnesene (32) and (6E)‐10,11‐dihydro‐nerolidol (33) were obtained (Scheme 8A, Tables S9 and S10, Figures S75–S90), demonstrating that SmTS6 cannot switch to a 1,6‐cyclisation mode. The absolute configuration of 33 was determined by chemical correlation to (R)‐ and (S)‐25. Catalytic hydrogenation of 33 and comparison to the 25 hydrogenation products 26 a and 26 b by GC using a chiral stationary phase confirmed the structure of (R)‐33 (92 % ee, Figure S91). Taking analogue 11 as substrate, two products (1E,5E,9E)‐1,5,9‐trimethylcyclododeca‐1,5,9‐triene (34, Table S11, Figures S92–S99) and (4E,8E)‐1,5,9‐trimethyl‐cyclododeca‐4,8‐dien‐1‐ol (35, Table S12, Figures S100–S107) were isolated. Their formation can be rationalised through a 1,12‐cyclisation of 11 to cation A5. Its deprotonation leads to 34, while attack by water gives raise to alcohol 35. Compound 34 is C 3h symmetric and shows only five signals in the 13C NMR. Its structure was confirmed by catalytic hydrogenation that resulted in the same products as obtained with 19 (Scheme 4, Figure S108).

Scheme 8

A) Enzymatic conversions FPP analogues 10 and 12 with SmTS6. B) Determination of the absolute configuration of 35 by labelling experiments with 36 and (R)‐ and (S)‐(1‐13C,1‐2H)IPP.

The absolute configuration of 35 was determined by labelling experiments with (R)‐ and (S)‐(1‐13C,1‐2H)IPP and the GPP analogue 36 that were converted into stereoselectively deuterated (R)‐ and (S)‐(1‐13C,1‐2H)‐11 with FPP synthase (FPPS) from Streptomyces coelicolor (Scheme 8B). The incorporation of labelling into 35 can be followed by HSQC spectroscopy into one of the diastereotopic positions assigned by NOESY spectroscopy with the unlabelled compound. Assuming inversion of configuration in the cyclisation to 35, the absolute configuration of (S)‐35 was concluded (Figure S109). With substrate 12, SmTS6 showed only poor conversion that was insufficient for product isolation, again in line with the inability of SmTS6 to perform 1,6‐cyclisations. For analogue 13, (S)‐23 was obtained, as confirmed by GC analysis on a chiral stationary phase (Figure S110).

Conclusion

Four synthetic FPP analogues were converted with the bacterial β‐himachalene synthase (HcS), (Z)‐γ‐bisabolene synthase (BbS), and germacrene A synthase (SmTS6), yielding interesting products in most cases. Most of these compounds are unknown or only poorly described. One exception is 34 that has been obtained before from isoprene in a metathesis reaction using a tungsten‐carbene complex, while the isomer 19 was only obtained as a mixture with 34 and other cyclotrimers of isoprene. Their enzymatic formation reported here opens the possibility to obtain both stereoisomers selectively with different STPSs. It is also interesting to note that different enzymes can produce enantiomers, as observed for the new compound 22 for which the enzymatic formation of its enantiomer was previously reported from the same FPP analogue. For compound 33 a synthesis from 6‐methylheptan‐2‐one has been reported. (E)‐iso‐γ‐Bisabolene, a stereoisomer of 30, has been tentatively identified by GC/MS in headspace extracts from the fungus Fusarium where it could be a side product of trichodiene synthase,[ , ] and the E/Z mixture of 30 has been obtained by synthesis, but the pure Z isomer has not been made available before. Also the enzyme products (R)‐ and (S)‐23, 24, 29, 31, 32 and 35 and their synthetic derivatives 27 and 28 were isolated in this study for the first time and obtained in high stereoisomeric purity, showing that the enzymatic conversion of FPP analogues can open the doors towards a new chemical space.

BbS catalyses an initial 1,6‐cyclisation of FPP and yielded corresponding products from all substrate analogues, only with substrate 13 for which a 1,6‐cyclisation is blocked the acyclic product (S)‐23 was obtained. The 1,10‐cyclising SmTS6 gave acyclic products with 10 for which such a cyclisation is not possible, however, substrate 13 that could in principle undergo 1,10‐cyclisation also only gave an acyclic product, likely because the conformational fold of this substrate in the active site is disturbed. With FPP analogue 11 a 1,12‐cyclisation was observed, demonstrating that new cyclisation paths can be opened that follow the intrinsic reactivity of the substrate. The results with HcS are less clear. With 10 no product was obtained, while with 11 a 1,12‐cyclisation and with 13 a 1,10‐cyclisation was manifested. However, these observations were contrasted by the conversion of 12 through initial 1,6‐cyclisation. We have discussed in our previous work that the HcS mechanism can be understood either by initial 1,6‐ or 1,11‐cyclisation. The results obtained here seem to indicate that HcS can switch between these modes, which may also be true for the natural FPP cyclisation. Future research may further clarify this point.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.