Germacrene A–A Central Intermediate in Sesquiterpene Biosynthesis
Graphical Abstract
Inspired by nature: This review discusses the unique chemistry of germacrene A, its occurrence in Nature and its central role as an intermediate in the biosynthesis of sesquiterpenes. Biosynthetic derivatives include eudesmanes and guaianes and further compounds arising from skeletal rearrangements. For each compound the reasoning for its assigned absolute configuration is discussed, if it has been established, and the natural sources are summarised.
Abstract
This review summarises known sesquiterpenes whose biosyntheses proceed through the intermediate germacrene A. First, the occurrence and biosynthesis of germacrene A in Nature and its peculiar chemistry will be highlighted, followed by a discussion of 6–6 and 5–7 bicyclic compounds and their more complex derivatives. For each compound the absolute configuration, if it is known, and the reasoning for its assignment is presented.
1 Introduction
With an estimated number of over 80,000 compounds terpenes form the largest class of natural products. They are produced by all kingdoms of life and can be classified as mono- (C10), sesqui- (C15) or diterpenes (C20) etc. according to the number of incorporated isoprenoid units. During the past decades many sesquiterpene synthases have been reported1-6 that catalyse the cyclisation of farnesyl diphosphate (FPP) through diphosphate abstraction to give the reactive farnesyl cation (A, Scheme 1). Attack of the C10=C11 double bond to C1 can yield the (E,E)-germacradienyl cation (B) by 1,10- or the (E,E)-humulyl cation (C) by 1,11-cyclisation. The alternative reaction by reattack of diphosphate to C3 results in nerolidyl diphosphate (NPP). After a conformational rearrangement of the vinyl group by rotation around the C2−C3 bond, cyclisation reactions may proceed to the (E,Z)-germacradienyl cation (D), the (E,Z)-humulyl cation (E), the bisabolyl cation (F), or to cation G, with possible formation of either enantiomer for chiral intermediates. Deprotonation of B leads to germacrene A, a widespread natural product and central intermediate in the biosynthesis of many 1,10-cyclised sesquiterpenes. This review discusses its occurrence in Nature, its chemistry, and central importance as an intermediate towards many sesquiterpenes.

Terpene cyclisation modes for FPP.
2 Germacrene A
2.1 Occurrence in Nature
(−)-Germacrene A (1, Scheme 2) was first isolated in 1970 from the gorgonian Eunicea mammosa.7 Its absolute configuration was established as (S)-(−)-1 through its Cope rearrangement to (+)-β-elemene (2) for which the configurational assignment was performed by chemical correlation of (−)-elemol (3) to (−)-2.8, 9 Compound (−)-1 is also believed to occur in the soft coral Lobophytum,10 and is the alarm pheromone of the aphid Terioaphis maculata.11, 12 In the course of this work it was noticed that the optical rotation ([α]D25=−26.8, c 1.0, CCl4) was significantly higher than initially reported ([α]D25=−3.2, c 14.4, CCl4),7 which is explainable by a partial rearrangement of purified (−)-1 to (+)-2, or alternatively, 1 isolated from E. mammosa was not enantiomerically pure. However, the optical rotation of (+)-2 ([α]D25=+15.1, neat) reported in this initial study7 matches the reported value for (−)-2 ([α]D25=−15.8, c 0.50, CHCl3) obtained by Cope rearrangement of (+)-1,13 thus disfavouring the latter hypothesis. In fact, the enantiomeric composition of a compound cannot be concluded only from the optical rotation upon its first isolation, or not with certainty if a compound is known to be instable. Methods such as chromatographic separation on a chiral stationary phase may be more conclusive. Through this approach, König and co-workers found that 1 from various plants is a mixture of enantiomers, ranging from nearly pure (+)-1 in Piper nigrum to mainly (−)-1 in the liverwort Barbilophozia barbata.14

Structure of 1 and its absolute configuration by chemical correlation.
Germacrene A synthase (GAS) catalyses a 1,10-cyclisation of FPP to B, followed by deprotonation to 1 (Scheme 3). Both enantiomers of 1 are accessible through this reaction, depending on whether C10 of FPP is attacked from the Re or the Si face. Since this face selectivity may be altered by subtle conformational changes of FPP in the active sites of GASs, predictions based on amino acid sequences or phylogenetic analyses regarding the stereochemical implications may be difficult. Many plant GAS have been identified during the past two decades, including two (+)-GASs from Cichorium intybus15, 16 and one from Matricaria recutita,17 with the absolute configuration of (+)-1 established by chiral GC. Sometimes the absolute configuration can be rationally suggested, because 1 is transformed in the same organism into another compound such as (+)-costunolide.18-21 Further GASs are known from many other plant species,22-32 but the absolute configuration of 1 has frequently not been determined. While the accumulated literature shows that (+)-1 is typical for plants, the recently characterised bacterial GAS from Micromonospora marina produces (−)-1,33 reflecting the observation that terpenes and cationic intermediates towards them from plants and bacteria often represent different enantiomers.34-37 The coinciding absolute configuration of (−)-1 from E. mammosa may point to a biosynthesis by symbiotic bacteria in the gorgonian.38

Cyclisation mechanism from FPP to A) (R)-(+)-1 and B) (S)-(−)-1.
2.2 Chemistry of germacrene A
The isolation and full structural and NMR-spectroscopic characterisation of 1 was a long-standing problem significantly hampered by its high reactivity. Its first isolation from E. mammosa in 1970 was done by extraction and concentration at temperatures below 35 °C to avoid the Cope rearrangement to 2 (Scheme 2).7 Chromatographic purification on slightly acidic silica gel induces a cyclisation through cation H1 to α-selinene (4), β-selinene (5), and selina-4,11-diene (6, Scheme 4).7, 11, 15

Acid catalysed conversion of 1 into selinenes.
The skeleton of 1 is characterised by a conformationally flexible 10-membered ring that shows sufficient ring strain to prevent a fast interconversion between conformers, resulting in broadened signals and multiple signal sets in the NMR spectra. Partial 1H- and 13C-NMR data were first published for 1 from T. maculata.12 Later studies improved the NMR data assignments for the main conformers of 1 (recorded at 25 °C), but did not allow for a completion of the data sets.13, 39 Through NOESY the conformers of 1 a (UU, Me14 and Me15 up), 1 b (UD, up-down) and 1 c (DU, down-up) in a 5:3:2 ratio were identified (Scheme 5 A).13 The NMR data sets (25 °C) for all three conformers were recently completed using a 13C-labelling strategy by conversion of all 15 isotopomers of (13C)FPP40 with GAS from M. marina into (−)-1, resulting in strongly enhanced 13C-NMR signals for the labelled carbons. HSQC spectroscopy of enzymatically prepared stereoselectively deuterated and 13C-labelled 1 allowed the NMR assignment of all hydrogens.33, 41

A) Conformers of (+)-1. B) Cope rearrangement of ent-1 a.
The stereoselectively deuterated and 13C-labelled isotopomers of 1 were also used to study the stereochemical course of its Cope rearrangement (Scheme 5 B). According to the Woodward–Hoffmann rules, pericyclic reactions follow a stereochemical course determined by the symmetry of frontier orbitals.42 For the Diels–Alder reaction this has been verified by stereoselective deuteration,43, 44 while classical experiments for the Cope rearrangement have been performed with meso- and rac-3,4-dimethylhexa-1,5-diene.45 The enzymatic access to labelled 1 allowed to follow the rearrangement to (+)-2 that proceeds from ent-1 a through a chair-chair transition state.33
For many terpene synthase reactions 1 is further cyclised in a second step initiated by reprotonation. This can occur at C1 and lead to the 6–6 bicyclic system of H as a precursor of eudesmane sesquiterpenes (Scheme 6 A). The 6–6 bicyclic system could in theory also arise by protonation at C4 leading to the secondary cation I, but this reaction is not preferred. Furthermore, 1 can be protonated at C10 with cyclisation to the 5–7 bicyclic skeleton of J, or at C4 resulting in K, representing the precursors to guaiane sesquiterpenes. As an alternative to the formation of neutral 1 and its reportonation also an intramolecular or water-mediated proton transfer in cation B may directly lead to H, J or K, thus bypassing 1 that would in such cases be better described as a side product rather than an intermediate. However, experimental evidence to distinguish between these alternatives is difficult to obtain, and 1 will preferentially be discussed as an intermediate towards more complex sesquiterpenes in this article. A detailed discussion of the reactions from 1 will follow in the subsequent sections.

Secondary terpene cyclisations of 1.
3 Eudesmanes
3.1 Eudesmanes with a regular skeleton
The protonation-induced cyclisation of 1 can lead to eight stereochemically distinct cationic intermediates (Scheme 7), four of which arise from (+)-1 (H1–H4), while the other four stereoisomers originate from (−)-1 (H5–H8). For each intermediate, simple deprotonations or nucleophilic attack of water are possible. Also, hydride shifts can occur first, which further widens the reachable chemical space of eudesmanes. For many of these possibilities the corresponding structures have been reported.

Cyclisations induced by reprotonation of 1 at C1 to H1–H8.
3.2 Eudesmanes from cation H1
An important intermediate to eudesmanes is H1. Deprotonations from C3 and C15 lead to α-selinene (4) and β-selinene (5), two compounds that have been isolated more than 100 years ago from celery oil.46 Their structures were elucidated in degradation experiments47 and were correlated to β-eudesmol (7, Scheme 8 A).48-50 Based on a comparison of physical characteristics of degradation products to those of other cis- and trans-decalins initially a cis-decalin structure was assigned,51 but a later conformational re-examination indicated a trans-fused ring system.52, 53 The absolute configurations of 4 and 5 were determined by chemical correlation through the following arguments. The structure of ketone 8 was established in the classical synthesis of steroids by Woodward.54 Two years later the same group converted 8 into the dicarboxylic acid 10 (Scheme 8 B) that was the opposite enantiomer as obtained by degradation of 7 (Schemes 8 C)55 that had previously been correlated with 4 and 5 (vide supra).

A) Structures of eudesmanes from H1 and of 7. B) Chemical correlation of ketone 8 with 10. C) Chemical correlation of 7 with ent-10.
The optical rotation of 4 was repeatedly found to have a positive value, including the reports from Brazilian rosewood oil ([α]D=+18),56 Dendropanax trifidus ([α]D=+68)57 and Cryptotaenia japonica ([a]D15=+6.3),58 or for 4 obtained by enantioselective synthesis ([α]D=+15.7, CHCl3).59 Andersen et al. pointed out that minor impurities may result in erroneous data and reported a value of [α]D=−16 (c 0.2, pentane)60 that was confirmed by Maurer and Grieder ([α]D20=−14.5, CHCl3, 1 %),61 and in both cases secured by CD spectroscopy. For 5 consistently positive optical rotations with values between [α]D=+31.7 (CHCl3) and [α]D=+60 (CHCl3) have been given.49, 57, 58, 60-66 Thus, natural α- and β-selinene from (+)-1 are characterised as (−)-4 and (+)-5. Complete 1H- and 13C-NMR data for 4 and 5 are available.66, 67
Compounds 4 and 5 were identified from various plant sources.49, 57, 58, 60-63, 66-76 In some cases 2 was also isolated,58, 68, 69 sometimes with determined absolute configuration of (−)-2,61-63 which supports (+)-1 as a biosynthetic intermediate, but 1 could also be the true natural product, while 4 and 5 may have been formed spontaneously from 1 during compound isolation (Scheme 4).
An alternative deprotonation of H1 can lead to selina-4,11-diene (6), while the attack of water may result in selin-11-en-4α-ol (12) or neointermedeol (13, Scheme 9 A). As the stereochemical information at C5 is lost in 6, this sesquiterpene can also arise from H4. Conclusions may be possible from co-isolated materials with retained stereochemical information at C5. The absolute configuration of 6 was evident from its formation by pyrolysis of the p-nitrobenzoate 14 of (−)-elemol (3), leading to (+)-6 (Scheme 9 B).65 This finding is further supported by an enantioselective synthesis of (+)-6 starting from (+)-trans-dihydrocarvone (15) through (+)-α-cyperone (16),77 followed by reduction of the ketone with AlCl2H (Scheme 9 C).78

A) Structures of eudesmanes from H1. B) Correlation of 14 to 6. C) Synthesis of 6.
Compound 6 has been isolated from several plants62, 74, 79-83 with reported positive optical rotations ranging from [α]D14=+32.05 (MeOH)79 to [α]D20=+54.5 (CHCl3, 1 %).80 From Vernonia glabra 6 was isolated together with 1, 2, 4 and 5 after column chromatography, suggesting that it may have been formed by silicic-acid-catalysed cyclisation of 1.81 The full61 or partial78, 79, 84 1H-NMR data have frequently been published, but unfortunately no 13C-NMR data are available from the literature.
The alcohol 12 ([α]D20=−18) was first isolated from Podocarpus dacrydioides and its structure was correlated to (+)-selinane (19), the hydrocarbon corresponding to 4 and 5, by catalytic hydrogenation to 17, dehydration with POCl3 to 18 and hydrogenation (Scheme 10 A), while the 4α orientation of the hydroxy function was deduced from the NMR spectrum, thereby establishing its absolute configuration.85 This structural assignment was confirmed by a synthesis from 7 that was converted into the epoxide and dehydrated with POCl3 to yield a mixture of 20 and 21 (Scheme 10 B). Epoxide opening with LiAlH4 resulted in (−)-12 and juniper camphor (22).86 Furthermore, the racemic compound, along with all other seven stereoisomers, has been synthesised87 and comparative spectroscopic data including 1H- and 13C-NMR have been published.87, 88 Identical 1H- and 13C-NMR data for 12 were reported for the material from Artemisia barrelieri89 and Tanacetum nubigenum.90 Compound 12 has been isolated from many plant species.73, 80, 82, 89-102

Chemical correlations of 12 with A) 19 and B) 7. C) Correlation of 13 with 6.
Neointermedeol (13) was first reported from the grass Bothriochloa intermedia, with an optical rotation of [α]D25= +7.5,103, 104 while the material isolated later from Panax ginseng exhibited a negative optical rotation ([α]D22=−4.8, c 3.45, CHCl3).84 To resolve the situation (−)-13 was dehydrated with POCl3 in pyridine, yielding (+)-6 and thus securing the absolute configuration of 13 (Scheme 10 C). The structure of 13 has also been confirmed by synthesis of the racemate.87 Further isolations have been reported from termites including Subulitermes baileyi105 and Amitermes excellens,106 and from the plants Geigeria burkei107 and Artemisia schmidtiana.108 Partial 1H- and full 13C-NMR data for 13 have been published.84, 104
3.3 Eudesmanes from cation H2
Sesquiterpenes arising through H2 occur less frequent in Nature compared to H1 derivatives, but the alcohol 26 (Scheme 11 A) is quite widespread. The sesquiterpene 5,10-diepi-α-selinene (23) was first reported from Dipterocarpus alatus ([α]D20=+2.1).109 The compound was co-isolated with (7R,10S)-eudesma-4,11-diene, (−)-25 ([α]D20=−108.6), that could potentially also arise by deprotonation of H3, but if a common terpene cyclisation is assumed, intermediate H2 should be relevant. The absolute configuration of 23 was assigned by epoxidation with peracetic acid to a mixture of stereoisomeric epoxides 28, reduction with LiAlH4 to yield a mixture of alcohols, and Jones oxidation. From the obtained ketones 29, the enantiomer of a known compound, was isolated as main product (Scheme 11 B).109 Further, an enantioselective synthesis of 23 from 30 that is readily accessible from dihydrocarvone 15 was reported, that proceeded by reduction with Li in NH3 and phosphorylation with (EtO)2POCl to 31, followed by defunctionalisation with Na in NH3 and tBuOH (Scheme 11 C).110 Alternatively, 30 can be converted into a mixture of 23, its C5 epimer and 25 by Wolff–Kishner reduction.111 The regioisomer 5,10-diepi-β-selinene (24) was first obtained along with 23 by dehydration of a sesquiterpene alcohol with the assigned structure of “paradisiol” (27) from grapefruit (Citrus paradisi).112 Subsequent work demonstrated that “paradisiol” was identical with intermedeol (26).113 All three compounds 23–25 were also obtained by hydrolysis of intermedeol β-d-fucopyranoside.114 Compound 23, sometimes accompanied by 24 or 25, has also been reported from several termites.106, 115, 116 Full 1H- and 13C-NMR data of 23 (with missing signals only for quaternary olefinic carbons) and 24 are available from the literature,116 while data for 25 are lacking.

A) Structures of 23–26 and “paradisiol” (27). Chemical correlations of 23 with B) ketone 29 and C) synthetic 30.
Intermedeol (4S,5S,7R,10S)-26 ([α]D25=+10.7) was first reported with 7S configuration from Bothriochloa intermedia.117 This wrong structural assignment was based on the finding that 26 was converted into (−)-selinane (19) by hydrogenation (Pd/C), dehydration (POCl3, pyridine) and hydrogenation (Pt/C, Scheme 12 A). The subsequently discovered alcohol 12 (Scheme 9)85 showed different physical characteristics and spectroscopic properties, and thus the structure of ent-12 for intermedeol was excluded. Oxidation of 26 with KMnO4 and NaIO4 to hydroxyketone 33, followed by epimerisation to 34 and Wittig methylenation gave ent-12, supporting a structural revision for intermedeol to 26. The initially observed formation of (−)-19 from 26 was explained by double bond migration and hydrogenation from the sterically less hindered side during Pd catalysis, yielding intermediate 32 with overall epimerisation at C7.86 The structure of 26 was also confirmed by synthesis.87, 110, 111

A) Chemical correlations of 26 with (−)-19 and ent-12, B) concerted mechanism for the protonation induced cyclisation of 1 to 26.
Compound 26 has frequently been isolated from plants.66, 117-127 For 26 isolated from Cymbopogon flexuosus the opposite absolute configuration was assigned, despite the optical activity of [α]D=+2 (c 3.3, MeOH). The compound was named “isointermedeol”,128 but this material was likely an impure sample of (+)-26.129 Nevertheless, the description of “isointermedeol” caused some confusion, as there is at least one later paper about Jasonia candicans with reference to the report of this supposedly new sesquiterpene alcohol.130 For the (+)-intermedeol synthase from Termitomyces GC-MS analysis of the products revealed minor amounts of 2, thereby establishing 1 as a side product and supporting this compound as a biosynthetic intermediate to 26.131 Another (+)-intermedeol synthase was recently reported from Streptomyces clavuligerus.132 Complete 1H- and 13C-NMR data of 26 in CDCl387, 88, 104, 114, 119 or C6D6131, 132 have been reported.
Paradisiol (4R,5S,7R,10S)-27 represents the initially assigned structure of a sesquiterpene alcohol from Citrus paradisi112 that was later corrected to 26.113 It may seem surprising that 27 has never been reported as a natural product, while its epimer 26 is widespread, but this is understandable on biosynthetic grounds (Scheme 12 B). Starting from the shown conformation of 1, a concerted protonation induced ring closure and attack of water can lead to 26, while the formation of 27 by such a process would require a syn addition to the C4=C5 double bond of 1 with attack of water from the internal face, which seems sterically impossible. However, compound 27 has been synthesised87 and was obtained as one of the hydrolysis products of intermedeol β-d-fucopyranoside ([α]D22=−17.9, c 0.53, EtOH).114 Full spectroscopic data are available.87, 88, 114
3.4 Eudesmanes from cation H3
Natural products from H3 are unknown. Synthetic compounds that could formally arise through H3 by terpene cyclisation include 10-epi-α-selinene (35), 7-epi-amiteol (36) and 5-epi-paradisiol (37, Scheme 13). Compound 35 was first obtained by Wolff–Kishner reduction of 30,111 and then from (R)-limonene (37) that can be converted in three steps into the aldehyde 39 (Scheme 13 B),133 followed by Wittig–Horner olefination to 40. An intramolecular Diels–Alder reaction results in the endo-adduct 4 and the exo-adduct 35 ([α]D25=+102, CHCl3, 0.7 %).134 A similar route was also reported from (S)-carvone.135 For 36 and 37 only synthetic routes to the racemates have been established.87 For all three compounds full spectroscopic data have been published.87, 88, 135

A) Structures of 35–37. B) Enantioselective synthesis of 35.
3.5 Eudesmanes from cation H4
Only a few natural products arising through H4 are known. Amiteol (+)-43 ([α]36524=+8, CHCl3) from the termite Amitermes excellens was the first isolated compound from this class and co-occurred with 5-epi-α-selinene (41), 5-epi-β-selinene (42) and 6 in this species (Scheme 14 A).107 Although 6 is usually assumed to be formed via H1, in A. excellens a formation via H4 is more likely, as this reflects the mechanism for its cometabolites. The absolute configuration of 43 was established by dehydration with SOCl2, yielding a mixture of 41, 42 and (+)-6 ([α]D24=+30, CHCl3),106 the same enantiomer as originally reported from Chamaecyparis formosensis.79 Furthermore, (+)-41 was synthesised from α-santonin (45) that was converted into 46 through a known route (Scheme 14 B).136 Reduction of 46 to epimeric diols 47, mesylation to 48 and elimination with Li2CO3 and LiBr in refluxing DMF yielded 41 ([α]D25=+30.1, c 3.50, CHCl3).137 Syntheses for racemic 43 and 5-epi-neointermedeol (44) have also been established,87 but despite its tentative GC/MS based identification as constituent of some essential oils compound 44 has not been isolated from natural sources so far. More recently, a terpene synthase for 41 has been identified from the cyanobacterium Nostoc punctiforme, but the absolute configuration of the product has not been assigned.138 Full spectroscopic data including IR, 1H- and 13C-NMR are available for 41,137, 138 43 and 44.87, 88

A) Structures of 41–44. B) Enantioselective synthesis of 41.
Notably, while the formation of the sesquiterpene hydrocarbons 41, 42 and 6 should be possible through H4, the formation of 43 along this pathway encounters a difficulty that is related to the explanation for the possible formation of 26, but not of 27, from H2 (Scheme 12). Along similar lines (Scheme 15 A), the protonation induced cyclisation of 1 starting from a boat–boat conformation can explain the biosynthesis of 44, while the formation of 43 would require the nucleophilic attack of water from the sterically less accessible Re face at C4. However, the formation of 43 is well understandable, if a precursor with a C4=C5 Z-configured double bond would be assumed (Scheme 15 B). This precursor is known as (−)-helminthogermacrene (49) from the fungus Helminthosporium sativum139 and later from the termite Amitermes wheeleri.140 The enantiomer (+)-49 was reported from the liverwort Scapania undulata and has a very similar EI mass spectrum and GC retention index to 1, but is less prone to a Cope rearrangement to (−)-cis-β-elemene (50, Scheme 15 C).141 Synthetic routes towards racemic 49 have been developed139, 142 and the absolute configuration of (+)-49 was established by chemical correlation to (−)-helmiscapene, a compound discussed in Section 3.8.39

Protonation induced cyclisations A) of 1 to 44 and B) of 49 to 43. C) Cope rearrangement of 49 to 50.
3.6 Eudesmanes from cation H5
Compounds derived from (−)-1 through the enantiomeric series of intermediates H5—H8 have been reported less often compared to those from (+)-1, which may be attributed to the fact that still most work has been done on higher plants for which (+)-1 is the typical enantiomer (Section 2). The cation H5 gives rise to the known natural products ent-α-selinene (ent-4), ent-β-selinene (ent-5), ent-selina-4,11-diene (ent-6) and (4S,5S,7S,10S)-eudes-11-en-4-ol (ent-12, Figure 1).

Structures of ent-4–ent-6 and ent-12.
The first report about naturally occurring enantiomers of selinane sesquiterpenes identified ent-4 as a constituent of the liverwort Chiloscyphus polyanthus in 1973. Its absolute configuration was established by CD spectroscopy in comparison to authentic (−)-4.60 Compounds ent-4 and ent-6, likewise established by CD spectroscopy and accompanied by 2, were subsequently reported from the liverworts Diplophyllum albicans and D. taxifolium,143 while the liverworts Riccardia jackii, Bazzania spiralis and Tylimanthus tenellus contain different combinations of ent-4, ent-5 and ent-12.144-147 Also insects were reported to contain ent-4 and (+)-2, exemplified by their occurrence in Ceroplastes ceriferus, which is surprising considering the fact that the „normal“ enantiomeric series of compounds is present in the related species C. rubens.62 In all these examples the absolute configurations were determined from the optical rotations of the isolated compounds. In Penicillium roqueforti also ent-4, ent-5 and ent-12 may occur; in this case the absolute configurations were assigned based on their biosynthetic relationship to aristolochene (vide infra) that is generated through (−)-1 in this fungus.148
3.7 Eudesmanes from cation H6
Little is known about eudesmanes arising via cationic intermediate H6. The compound 7-epi-α-selinene (ent-23, Scheme 16 A) was first reported from Amyris balsamifera, a species from which also 7-epi-α-eudesmol (51, Scheme 16 B) was isolated and structurally characterised by NMR spectroscopy. From its positive optical rotation ([α]D=+10, c 1.8, CHCl3) the authors concluded on the shown absolute configuration for 51, but a comprehensible explanation for this assignment is missing. Dehydration of 51 yielded a mixture of two products to which the structures of ent-23 and 52 were assigned by NMR spectroscopy, unfortunately without separating the obtained materials and determining their optical rotations. The compounds described as ent-23 and 52 also occurred in the essential oil of A. balsamifera.149 One study reported the chromatographic separation of the compound from A. balsamifera and (+)-23 (the latter with a mentioned source „provided by Dr. Wilfried König“) on a chiral stationary GC phase, which represents the only hint in the literature that the structure of ent-23 for the essential oil constituent may be correctly assigned.150 Compound ent-23 was also reported as major product of a terpene synthase from Vitis vinifera.150, 151 Both enantiomers of 23 have been obtained by synthesis from the enantiomers of 15, but optical rotary powers of the products were not measured.152 However, ent-23 may have a negative optical rotation, as for 23 from Dipterocarpus alatus a low value of [α]D20=+2.1 was determined.109 This would be consistent with a report by König in which ent-23 was published as the (−)-enantiomer, albeit only based on separation by gas chromatography using a chiral stationary phase without isolation.153

A) Structures of ent-23, ent-25 and ent-26. B) Dehydration of 51.
Compound ent-25 ([α]D16=+46.5, c 0.85, CHCl3) has been synthesised using the same strategy as for 6 (Scheme 9 C),78 but has not been isolated from any organism. The only report about ent-26 from Monactis macbridei by Bohlmann and co-workers154 gives a reference to the erroneous “isointermedeol”128 that was corrected shortly after.129 Unfortunately, Bohlmann's paper does not give an optical rotation for the isolated material so that it is difficult to judge, if the authors of this study were aware of the misassignment of “isointermedeol” at the time of their publication. Overall, this discussion shows that compounds from H6 are not only rare, but if they occur in the literature, the assignments of absolute configurations remain unclear. Since the compounds originate in all cases from higher plants, they may truly be the usual enantiomers, that is, 23, 25 and 26.
3.8 Eudesmanes from cations H7 and H8
The literature contains only few reports of compounds that may originate from H7, while no examples from H8 are available. α-Helmiscapene (ent-35, Scheme 17 A) was first isolated from Scapania undulata and suggested to arise through a “cis-germacrene”,155 a compound that was later described from this species141 after its first identification from H. sativum as helminthogermacrene (49).139 In agreement with the positive optical rotation of synthetic 35 (Scheme 13), ent-35 was found to be the (−)-enantiomer ([α]D=−100, CHCl3) and correlated to (+)-δ-selinene (54) by acid-catalysed isomerisation (Scheme 17 B). Both ent-35 and β-helmiscapene (−)-53 were also found in the liverwort Radula perrottetii.156 The acid-catalysed cyclisation of (+)-49 to ent-35 suggests that the formation of ent-35 from 49 could be non-enzymatic and that germacrene A may indeed not be the precursor of helmiscapenes.39 Full 1H- and 13C-NMR data are available for ent-35 and 53.39, 156

A) Structures of ent-35 and 53. B) Acid-catalysed cyclisation of 49 to ent-35 and isomerisation to 54.
4 Rearranged Eudesmanes
In this section rearranged eudesmanes from H1–H6 will be discussed, while such compounds from H7 and H8 are unknown.
4.1 Rearranged eudesmanes from H1
Rearranged eudesmanes can in theory arise from all cations H1–H8 in Scheme 7. An important group of compounds by widespread occurrence in Nature originates from H1. Specifically, this intermediate can undergo a 1,2-hydride migration to H1 a that must proceed suprafacially and thus determines the configuration at C4 (Scheme 18; 1,n-hydride or proton migrations as used in this article refer to the distance of n carbons for the migration, not to positional numbers). A subsequent 1,2-methyl group migration leads to H1 b (path a) that upon deprotonation yields eremophilene (55) or 4,5-diepi-aristolochene (56). Alternatively, H1 a can react in a Wagner–Meerwein rearrangement (WMR) with ring contraction to H1 c that results in hinesene (59, path b).

Biosynthesis of rearranged eudesmanes from H1.
Compound 55 was first isolated from Petasites officinalis and P. albus ([α]D20=−104.2 and [α]D24=−142.5, respectively).157-159 Its structure was initially wrongly assigned,160 but then corrected based on a chemical derivatisation and interpretation of the EI-MS fragmentation behaviour of a thioketal derivative.159 The sesquiterpene 55 was later isolated from several higher plants.58, 161-167 Furthermore, (−)-55 was discovered in the gorgonian Plexaurella fusifera168 and along with 2 in the liverwort Frullania serratta.169
An elegant synthesis for (rac)-55 has been developed starting from 60 that can give 61 a by a Diels–Alder reaction, with partial epimerisation to 61 b (Scheme 19 A). Both compounds can be converted into 62 by acid-catalysed isomerisation. Reaction with tosylhydrazine leads to 63 that was reduced with NaBH4 via 64 to 65.170 Treatment with MeLi and dehydration with SOCl2 in pyridine gave 55.171 Its double bond regioisomer 56 (Scheme 19 B) was first obtained from eremophilone (66), the first structurally characterised terpene found to violate Ruzicka's isoprene rule,172 by reduction with LiAlH4 and AlCl3,173 and later from eremophil-9-en-11-ol (67) by dehydration ([α]D=−11.1, c 0.18, CHCl3).174 Compound 56 has also been obtained by synthesis from capsidiol,175 but was never isolated from Nature. Complete 13C-NMR data are available for 55 and 56.170, 175

A) Synthesis of (rac)-55 through a Diels–Alder approach, B) preparation of 56 from the natural products 66 and 67.
The sesquiterpene alcohol 4αH-eudesma-11-en-4α-ol (57), [α]D=+32.8 (c 0.7, CHCl3), was isolated from Kleinia pendula and can arise by attack of water to H1 a.176 Similarly, the addition of water to H1 b leads to eremophil-11-en-10β-ol (58), a compound that is known from Alpinia intermedia ([α]D= +29.2, c 0.12, CHCl3).66 For both alcohols 57 and 58 full 13C-NMR data were given.66
Hinesene (59) was first isolated from Rolandra fruticosa ([α]D24=−44, c 0.1, CHCl3).177 The absolute configuration was initially assigned based on the same sign of optical rotation than for hinesol and later confirmed by enantioselective synthesis from santonin.178 The compound is also known from an unspecified liverwort of the genus Frullania.179 Full 1H- and 13C-NMR data were provided.177, 178
4.2 Rearranged eudesmanes from H2
Also rearranged eudesmanes from H2 constitute an important group of compounds (Scheme 20 A), including (+)-valencene (68), (−)-aristolochene (70), valencene hydrate (71) and its C10 epimer 72, (−)-ishwarane (73), (−)-8,12-seco-ishwaran-12-ol (74) and (−)-agarospirene (71). Compound 73 requires a third cyclisation from H2 b to H2 c and deprotonation with closure of a cyclopropane ring, while 74 can be explained by attack of water to H2 c.

A) Biosynthesis of compounds from H2. B) Dehydration of 76 to 68.
Valencene (68) was first isolated from orange oil180 and found to be related to nootkatone (69) by oxidative conversion,181 an important value adding transformation for which an artificial enzyme system has been developed.182 Compound 69 is a flavour constituent of citrus fruits and its structure had previously been established.183 The optical rotation of 68 was determined for the material obtained by dehydration of valerianol (76, Scheme 20 B) with NaOAc in refluxing Ac2O ([α]D=+73.4, c 5.3, CHCl3).184 A synthesis of (rac)-68 similar to the synthesis of (rac)-55 in Scheme 19 A has been developed.170 The sesquiterpene 68 is a constituent of the essential oils from numerous plants, but has rarely been isolated. Bixa orellana is one of the few sources from which its isolation was mentioned,185 while it was obtained enriched together with 2 in a sesquiterpene hydrocarbon fraction from the liverwort Porella acutifolia.186 The combination of 2 and 68 also occurs in the octocoral Plexaurella fusifera,168 while 68 from bacteria is rare, but has been identified from Streptomyces sp. FORM5.187 Valencene synthases are known from Citrus sinensis,188 Vitis vinifera,150, 151 and Callitropsis nootkatensis,189 in which it occurs together with a valencene oxidase for the biosynthesis of 69.190 Besides 68, the terpene synthases from V. vinifera were reported to produce (−)-7-epi-selinene (ent-23, Scheme 16)150, 151 that must originate from H6. It would be easier to understand, if one of the two enzyme products would represent the opposite enantiomer than reported, so that both could arise through a common intermediate. In fact, the configurational assignment for 68 was based on a GC analysis using a chiral stationary phase, but without including a (−)-68 standard.
Aristolochene (70, [α]D25=−76.47) was first isolated from Aristolochia indica. Its structure was elucidated by NMR spectroscopy and catalytic hydrogenation, yielding a mixture of (+)-nootkatane (77), also obtained by hydrogenation of 68, and its C10 epimer 78 (Scheme 21 A).191 The structural assignment was later confirmed by a synthesis of 70 from 68, that was first oxidised to 69, followed by conversion into the dienol acetate 79 (Scheme 21 B). Deconjugation by reduction with NaBH4 gave 80 that was defunctionalised with thiocarbonyldiimidazole and Bu3SnH to yield 70.192 Furthermore, an enantioselective synthesis from (S)-carvone (81) has been developed (Scheme 21 C). After silylation to 82, a Robinson annelation with ethylvinyl ketone resulted in 83. Its reduction with excess LiAlH4 and AlCl3 to 84 was followed by epoxidation to 85. Treatment with TiF4 resulted in epoxide opening with methyl group migration and cleavage of the trimethylsilyl cation to produce 86, that was defunctionalised in two more steps to 70.193 Compound 70 was also reported as a side product of valencene synthase from V. vinifera150 and as a headspace constituent from Streptomyces acidiscabies.194 Both compounds (+)-68 and (−)-70 are present in extracts from the liverwort Dumortiera hirsuta with absolute configurations established in comparison to authentic standards by GC using a chiral stationary phase.153 Full 13C-NMR data for 68170 and 70193, 195, 196 have been reported.

A) Hydrogenation of 70. B) Synthesis of 70 from 69, and C) from (S)-carvone (81).
Valencene hydrate (71), arising from H2 b by attack of water, has been isolated from orange juice. For comparison this compound and its C10 epimer 72 were synthesised from 68 by epoxidation and epoxide opening with LiAlH4. Unfortunately, no optical rotations were given, but full 13C-NMR data are available.197
(−)-Ishwarane (73, [α]D=−40.33) was first isolated from Aristolochia indica where it co-occurs with biosynthetically linked 70.191 The compound has been chemically correlated through (+)-ishwarone (87) that can be converted into 73 by Wolff–Kishner reduction (Scheme 22).191 Compound 87 undergoes ring opening to (−)-isoishwarone (88) by treatment with acid.198 Its further conversion by acetalisation, hydroboration and oxidation leads to 89, that upon deacetalisation and retro-aldol reaction results in 90. Reduction through the bis-semicarbazone yields (+)-nootkatane (77), thus firmly establishing the absolute configuration of 73.199 Ishwarane was subsequently also found in many other plants,185, 200-204 while 8,12-seco-ishwaran-12-ol (74, [α]D=−165, c 0.1, CHCl3) has only once been reported from Litsea amara.205 Its absolute configuration has not been formally established, but was suggested to correspond to that of 73. Full 13C-NMR data for 73 and 74 are available.206, 207

Chemical correlation of ishwarane (73) with nootkatane (77).
(−)-Agarospirene (75) was first obtained by pyrolysis of the benzoate ester of agarospirol, a compound isolated from agarwood.207 Its structure has also been ascribed to a natural product isolated from the liverworts Scapania robusta and Scapania maxima,208, 209 but a later synthesis of 75 ([α]D22=−11, c 0.3) and its stereoisomers demonstrated that the natural product was identical to (−)-hinesene (59).178 Complete 1H- and 13C-NMR data for 75 were reported.178
4.3 Rearranged eudesmanes from H3
Natural rearranged eudesmanes from H3 are unknown. The only known compound is (4S,5R,7R)-spirovetivadiene (91) that has been obtained by synthesis ([α]D22=−3, c 0.6). Its hypothetical biosynthesis from H3 would require a 1,2-hydride shift to H3 a, ring contraction to H3 b and deprotonation (Scheme 23). Full 1H- and 13C-NMR data are available.178

Rearranged eudesmanes from H3: spirovetivadiene (91).
4.4 Rearranged eudesmanes from H4
Known rearranged eudesmanes from intermediate H4 (Scheme 24) are represented by (−)-4-epi-eremophilene (92), (+)-5-epi-aristolochene (93), (−)-premnaspirodiene (95, also named spirovetivene), (−)-spirolepechinene (96) and 4βH,7αH,10β-eudesm-11-en-4α-ol (98). The unusual sesquiterpene 97 requires a ring contraction to H4 d and deprotonation.

Biosynthesis of rearranged eudesmanes from H4.
Both compounds 92 ([α]D25=−22.7, c 0.17, CHCl3) and 93 ([α]D25=+8.13, c 0.16, hexane) were obtained by synthesis from capsidiol (94).175, 210 Notably, 93 is also the biosynthetic precursor to 94,211 as was demonstrated by incubation of [1,1-3H2]FPP with cell-free enzyme preparations from Nicotiana tabacum, yielding radioactively labelled 93. Furthermore, 14C-labelled 93 was incorporated into 94 in feeding experiments with N. tabacum and Capsicum annuum.212, 213 Subsequent work resulted in the purification of tobacco 5-epi-aristolochene synthase (TEAS),214 cloning of the genes from N. tabacum and C. annuum and expression in Escherichia coli,215-217 and determination of the first crystal structure of a plant terpene synthase.218 Based on this structure the active site residue Tyr520 was suggested to be responsible for reprotonation of the intermediate (−)-1. Consistent with this hypothesis, the Y520F enzyme variant gave (−)-1 as a single product.219 Also the 5-epi-aristolochene-1,3-dihydroxylase for the biosynthesis of 94 from 93 has been identified.220 For the biotechnological access to 93 the epi-aristolochene synthase gene has been heterologously expressed in E. coli,221 in Oryza sativa,222 and in yeast in which optimisation of the strain and the culture conditions resulted in a high titre production.223 A thermostable variant of EAS has been created.224
Along similar lines of research, 95 has first been isolated from Premna latifolia225 and subsequently from Lepechinia bullata ([α]D20=−88, c 0.501, CHCl3) in which it co-occurs with 97 ([α]D20=−32, c 0.125, CHCl3).226 The premnaspirodiene synthase (also known as vetispirodiene synthase) from Hyoscyamus muticus (HPS) has been characterised.227, 228 Another sesquiterpene synthase (Tps32) from Solanum lycopersicum with 90 % sequence identity to HPS was initially described as viridiflorene synthase,229 but a later study showed that Tps32 is indeed active as premnospiradiene synthase.230 Compound 95 is the parent hydrocarbon of (−)-solavetivone (96),231, 232 for which a premnaspirodiene oxygenase was reported.233
A detailed analysis of the product profiles of TEAS and HPS has led to the characterisation of several side products and demonstrated that TEAS produces minor amounts of 95,234 while HPS generates small quantities of 93 from FPP.235 Domain swapping experiments between TEAS and HPS resulted in enzyme variants making mixtures of 93 and 95 and allowed the identification of domains that conferred specificity for these two products.236 After the crystal structure of TEAS had become available, a systematic and rational approach targeting nine selected residues within and near the active site in all 29=512 combinations for a functional interconversion between TEAS and HPS was surveyed.237, 238 Finally, compound 98 has been isolated from orange juice. 1H- and 13C-NMR data for 92,175 93,210 95,178, 226 97,226 and 98197 have been published.
4.5 Rearranged eudesmanes from H5
Only a few reports about rearranged eudesmanes from H5 from Nature are available (Scheme 25). Terpene synthases for ent-55 have been characterised from the myxobacterium Sorangium cellulosum ([α]D25=+131.7, c 1.0, CHCl3)239 and the plant pathogenic fungus Fusarium fujikuroi.240 The cyclisation mechanism of (+)-eremophilene synthase from F. fujikuroi was studied by isotopic labelling experiments that showed selective deprotonation from C12 of FPP in the formation of the intermediate (−)-1, allowed to follow the 1,2-hydride shift from H5 to H5 a, and demonstrated that the final deprotonation from H5 b to ent-55 proceeds with loss of the same proton as incorporated in the cyclisation of (−)-1 to H5 (Scheme 7).240 A crystal structure of ent-55239 and full NMR data assignments have been published.239, 240 Only a synthetic study towards ent-56 ([α]D25=+12.5, c 2.5, CHCl3) is available.241

Biosynthesis of rearranged eudesmanes from H5.
4.6 Rearranged eudesmanes from H6
Rearranged molecules from H6 (Scheme 26 A) are (−)-valencene (ent-68) and (+)-aristolochene (ent-70) that has been isolated from Aspergillus terreus ([α]D=+79.4, c 0.0176, hexane),192, 196 and Penicillium roqueforti, in which it occurs together with 2.148, 242, 243 The absolute configuration has been established by synthesis of (−)-70 from (+)-valencene (68).192 (+)-Aristolochene synthase was first isolated from P. roqueforti (PR-AS)244 and is also present in A. terreus (AT-AS).245 Subsequent gene cloning and expression gave efficient access to the recombinant enzymes.246, 247 A biphasic flow reactor system for the biocatalytic production of ent-70 has been developed.248

A) Biosynthesis of rearranged eudesmanes from H6. B) Cyclisation of (R)-5,6-dihydro-FPP (100) to 101 by AT-AS. C) Proposed water-mediated proton transfer from (S)-B to M in the biosynthesis of ent-70.
Notably, PR-AS produces a mixture of ent-70 as the main and ent-68 and (−)-1 as side products, while AT-AS yields ent-70 as a single product.249, 250 Isotopic labelling experiments demonstrated that the cyclisation of FPP to ent-70 proceeds with inversion of configuration at C1 and the specific loss of a proton from C12.245 The E252Q variant of PR-AS yielded (−)-germacrene A (1) as the only product.250 Further support of (−)-1 as an intermediate was obtained by the observed cyclisation of (R)-5,6-dihydro-FPP (100) to the germacrene A analogue 101 by AT-AS (Scheme 26 B).251 Similar experiments have been carried out with fluorinated FPP analogues.252, 253 On the other hand, instead of a true pathway intermediate, (−)-1 could only be a shunt product. Allemann and co-workers have argued for this view, as (−)-1 was not accepted as a substrate by PR-AS,249 and a computational study showed feasibility of a water-mediated direct proton transfer from (S)-B to M that could further cyclise to H6 (Scheme 26 C).254 However, the same workers later excluded this possibility experimentally, because the incorporation of deuterium from D2O at C1 of ent-70 proceeded with Re face attack.255 Based on the crystal structure of PR-AS the active site residue Tyr92 was suggested to serve as a general acid in the reprotonation of (−)-1,256 but also this hypothesis was disfavoured by site-directed mutagenesis.250 A more detailed picture was subsequently obtained by the crystal structure of AT-AS, providing evidence that the diphosphate anion is ideally positioned to act as a general acid and base relevant for i) the deprotonation of (S)-B, with the proton taken up by O6, and ii) the reprotonation of the resulting (−)-1 with donation of a different proton from O3 (this process may also be concerted with 1 as a highly transient species, Scheme 26 D).257 The results of a site-directed mutagenesis suggest that the thus formed eudesmane cation H6 is stabilised by W334 of PR-AS or W308 of AT-AS.258 Cationic aza-analogues of H6 have been shown to efficiently inhibit catalysis by PR-AS.259, 260
The sesquiterpene hydrocarbon ent-70 is the biosynthetic precursor to PR toxin (99),261 a potent mycotoxin that targets transcription and protein biosynthesis with a lethal dose of LD50=5 mg kg−1 in mice,262-264 and a series of other oxidation products that are likely pathway intermediates.265-269 Surprisingly, despite the potential of mycotoxin biosynthesis P. roqueforti is traditionally used for the production of blue cheese, which is explainable by the rapid degradation of 99 under cheese fermentation conditions.270 Biosynthetic hypotheses linking these oxidised metabolites have been investigated by feeding of labelled precursors148, 269 and discussed on the grounds of the biosynthetic gene cluster,271-273 but apart from the aristolochene synthase and the poorly characterised eremofortin C oxidase274 for the installation of the aldehyde function in 99 little is known about the enzymes involved in fungal toxin biosynthesis.
5 Guaianes
5.1 Guaianes formed by C4 protonation of germacrene A
Eight cationic intermediates can be formed from the enantiomers of 1 by protonation at C4 and ring closure (Scheme 27). These cations exhibit four stereogenic centres, leading to a maximum number of 24=16 possible stereoisomers, but two of the stereogenic centres are not set independently, since the C4/C5 double bond in 1 is E-configured and the ring closure proceeds by anti addition, that is, Me15 and H5 must be arranged trans. Thus, only eight stereoisomers are relevant to this pathway, namely J1–J4 from (+)-1, and their enantiomers J5–J8 from (−)-1.

Cyclisations induced by reprotonation of 1 at C4 to J1–J8.
5.2 Guaianes formed from cations J1 and J2
Guaianes from cations J1 and J2 include δ-guaiene (102) and pogostol (103, Scheme 28 A). δ-Guaiene is also named α-bulnesene and can in principle be generated by the deprotonation of J1 or J2, while 103 derives from J2 by Si face attack of water. Compound 102 was first isolated from the patchouli oil of Pogostemon cablin and given its premier name δ-guaiene in 1950. Initially, only the planar structure with insecure positioning of double bonds was determined, with a reported optical rotation close to zero of [α]D=+0.32.275 Later, bulnesol (107) was chemically converted into 102 by pyrolysis of its acetate 108 (Scheme 28 B), leading to a material with an [α]D=0,276 that was thus inconclusive for assigning the absolute configuration of 102 from the fully established structure of 106.277, 278 Because 102 is accompanied by patchouli alcohol (106) in P. cablin, it was suggested that both compounds should have coinciding absolute configurations, but at this time for 106 still a wrong structure was assumed (vide infra).276 A subsequent stereoselective synthesis from α-cyperone (16, Scheme 9) and comparison of the optical rotatory dispersion (o.r.d.) curves of synthetic and natural 102 finally established its structure.279, 280 Compound 102 is known from several other plants281-284 including Piper fimbriulatum,285 in which it occurs together with 2. In addition, 102 can be produced by cultured cells from Aquilaria crassna and Aquilaria sinensis,286, 287 resulting in the discovery of the δ-guaiene synthase from A. crassna.288 Compound 102 is also one of the main products of the α-guaiene synthase from V. vinifera289 and a side product of the patchoulol synthase from P. cablin.22, 290 The complete 1H- and 13C-NMR data of 102 are available.286

A) Guaianes derived from J1 and J2, initially reported structures of pogostol (104) and pogostol methyl ether (105), and patchoulol (106). B) Synthesis of 102 from bulnesol (107).
Pogostol (103) was first isolated from P. cablin ([α]D=−20.2, c 8.7).291 Since then, 103 was reported from various other plant sources292-296 and is known from the fungus Geniculosporium.297 A relative configuration was first assigned for pogostol O-methyl ether (105) from Artabotrys stenopetalus,298 followed by the assignment of the relative configuration of 104 for pogostol by Weyerstahl and co-workers.293 A subsequent synthesis of the reported structures 104 and 105 for pogostol and its methyl ether demonstrated that both assignments were erroneous.299 Amand et al. then gave a correction as 103.295 Although pogostol is long known and fairly widespread in Nature, the absolute configuration still remains to be determined. For unclear reasons the structure of ent-103 has been assigned to the CAS number of pogostol (21698-41-9), while in fact 103 may be more likely, because this corresponds to the main product 106 of the patchoulol synthase from P. cablin that also makes 103 as a side product.22 1H- and 13C-NMR data of 103 are reported in the literature.292-295, 297
The sesquiterpene 1,4-diepi-γ-gurjunene (109, Scheme 29 A) was isolated from the sponge Cymbastela hooperi ([α]D= +34.6, c 0.11, CHCl3).300 The formation of this compound can be understood from J1 by two sequential 1,2-hydride shifts via J1 a to J1 b and deprotonation. Since the absolute configuration of 109 has not been determined, it may also be derived from intermediate J5. Full 1H- and 13C-NMR data have been provided for 109.300

Biosynthesis of guaianes from A) J1 and B) J2. C) Chemical correlation of 110 with guaiol (114).
α-Guaiene (110, Scheme 29 B) may instead arise from J2 by 1,2-hydride migration to J2 a and deprotonation. It is the universal precursor leading under simple aerial oxidation conditions to many fragrant volatiles of industrial importance such as (R)- and (S)-rotundols (111 and 112) and rotundone (113) that exhibit a pleasant peppery or woody aroma.301-303 Compound 110 ([α]D19=−64.5, c 3.584, dioxane) was initially obtained by dehydration of guaiol (114, Scheme 29 C).304 With the absolute configuration of 114 being specified,305 the full structure of compound 110 was also affirmed. Natural sources of 110 include several plant species63, 284, 285, 306-310 and cell cultures from Aquilaria crassna and A. sinensis.286, 287 A recombinant α-guaiene synthase has been reported from V. vinifera,289 and 110 is also a side product of δ-guaiene synthase from A. crassa288 and patchoulol synthase from P. cablin.22, 290 The biosynthesis of 110 is also possible from K1 (Scheme 32, Section 5.5) by 1,2-hydride shift and deprotonation, but the co-occurrence with 102 in several species,284-287, 307, 308 whose formation can best be understood from J1 or J2, together with the observation of both compounds in the product profiles of several terpene synthases22, 288-290 speaks in favour of a common biosynthesis through J2. Full 1H- and 13C-NMR data of 110 are provided.284, 308
5.3 Guaianes formed from cations J3 and J4
Guaianes from J3 and J4 include guaia-1(10),11-diene (115) that is accessible through both cations by deprotonation, and guaia-9,11-diene (116) obtainable by loss of a proton from J3 (Scheme 30 A). Deprotonation of J4 can lead to guaia-10(14),11-diene (117), a compound for which we revise the structure here based on the reason given below, while the attack of water to J4 can give 4,5-diepi-pogostol (118). For 118 this discussion is hypothetical, because this compound was only obtained in racemic form by synthesis and is not known as natural product.299

A) Structures of 115–118. Correlations through hydrogenation products B) of 115 and 116 to 120 and C) of revised 117 to aciphyllene (122, see text).
The hydrocarbons (+)-115 and (+)-116 were both isolated only from the fruits of Peucedanum tauricum.311 Their co-occurrence in one organism suggests that they may have the same cationic precursor J3. The absolute configurations of 115 and 116 were specified by comparison of their hydrogenation products to those obtained from (+)-γ-gurjunene (120, Scheme 30 B),312 leading to one common product (119 a) from all three materials, as judged by GC analysis using two different chiral stationary phases.
Guaia-10(14),11-diene (117) is only known from Abies koreana.121 Its absolute configuration was elaborated using the same hydrogenation strategy as for 115 and 116 with chemical correlation to aciphyllene (122, Scheme 30 C). At the stage of this work the structure of 123 with 7S stereochemistry was assigned for aciphyllene,284 which would have led to the hydrogenation products 119 f and 119 i, and therefore the structure of 121 was concluded for the natural product from A. koreana expected to give the hydrogenation products 119 f and 119 g. However, shortly after the structure of aciphyllene underwent a revision to (7R)-122.313 In conclusion, the truly obtained hydrogenation products from aciphyllene were 119 c and 119 h, with the consequence that the natural product from A. koreana must be revised herewith to 117, expected to give 119 c and 119 d.
The synthetic compound 1-epi-aciphyllene (124) has been prepared from guaiol (114),314 but has not been discovered from Nature so far. Indeed, its biosynthesis is not easily understood, as its formation through the K series (Scheme 32, Secion 5.5) of cations cannot lead to a cis-orientation of H1 and Me14. If 124 exists at all as a natural product, two sequential 1,2-hydride migrations from J4 to J4 a and deprotonation could explain its formation (Scheme 31). Full 1H- and 13C-NMR data for 124 were reported,314 but unfortunately no optical rotation that would be useful for comparison in case of its future isolation.

Hypothetical biosynthesis of 1-epi-aciphyllene (124).
5.4 Guaianes formed from cations J5–J8
Despite the fact that for 103 the absolute configuration has not been determined and this compound could in principle arise through J6, no guaianes from J5–J8 are known. The absolute configuration of 1,4-diepi-γ-gurjunene (109) from C. hooperi would be most interesting to know, as sponges may produce the optical antipodes of plant compounds.
5.5 Guaianes formed by C10 protonation of germacrene A
Considering the discussion above, there are also only four logical cationic intermediates (K1–K4) after the cyclisation from (+)-1 initiated by C10 protonation (Scheme 32). Likewise, (−)-1 can produce four additional candidates (K5–K8).

Cyclisations induced by reprotonation of 1 at C10 to K1–K8.
5.6 Guaianes formed from cations K1 and K2
A deprotonation from C5 of K1 or K2 provides aciphyllene (122), also named guaia-4,11-diene. Compound 122 was first isolated from Lindera glauca in 1983 ([α]D20=+153.0).284 Its structure was erroneously elucidated by Kubota et al. as that of 7-epi-aciphyllene (123) by chemical correlation with aciphyllic acid (125, Scheme 33),284, 315 a compound that had been reported with 7S configuration.316 The structure was later corrected to 122 by synthesis from (+)-dihydrocarvone (15).313 Whether this means that also 125 should be revised to have 7R configuration or the material had undergone epimerisation at C7 during the transformations into 122 remains unclear at this stage. However, since Kubota and co-workers315 as well as Liu and Yu317 have reported different NMR data for “aciphyllic acid”, in both cases with 7S configuration, at least one of these structures must be wrong. Thus it may be likely that the Japanese workers have indeed started their correlation of “aciphyllic acid” to 122 from a material with 7R configuration. (+)-Aciphyllene (122) was later also found in Dumortiera hirusta,153 and with undetermined absolute configuration from the essential oil of Xylopia rubescens.310 It is also known as a side product of the recombinant patchoulol synthase from Pogostemon cablin,290 a multi-product terpene synthase for which all products retain the (7R) stereochemistry introduced in the intermediate (+)-1 and thus further supporting the structural reassignment for 122. Moreover, total syntheses from (R)-limonene by Srikrishna et al.318 and from guaiol (114) by Huang et al.314 were conducted. The 1H- and 13C-NMR data of 122 have been published.153, 284

Chemical correlation of “ciphyllic acid” to 122 (corrected structure).
5.7 Guaianes formed from cations K3 and K4
One of the most important sesquiterpenes derived from the K series is (+)-γ-gurjunene (120). Its formation can be understood from K4 by 1,2-hydride shift to K4 a and deprotonation (Scheme 34 A). This component was first discovered from the gurjun balsams of several species of Dipterocarpus ([α]D= +147, CHCl3).314, 319 Its absolute configuration was illuminated by correlation with α-gurjunene (127) and guaiol (114, Scheme 34 B).312 While treatment of 127 with acid gave the isomerisation products (+)-128 and 120 identical to natural (+)-γ-gurjunene, the isomerisation of 114 produced (−)-ent-128. Compound 120 was also isolated from Persea gamblei.320 Complete 1H- and 13C-NMR data have been published.300, 319, 321

A) Biosynthesis of 120. B) Correlation of 120 with 127 and 114.
Compound (−)-ent-123 (Figure 2) is only known as a synthetic material ([α]D24=−13.2, c 0.35, CHCl3) and could, as a hypothetical natural product, arise from K3 or K4 by deprotonation. It is wrongly presented in the synthesis paper that corrects the structure of (+)-aciphyllene (122) as the assigned structure of this natural product (123, Scheme 30), while it represents in fact its enantiomer. Full 1H- and 13C-NMR data are available.313

Structures of synthetic compounds ent-123, 123 and ent-120.
5.8 Guaianes formed from cations K5–K8
Natural products from the cations K5–K8 are unknown. Synthetic compounds (Figure 2) include (+)-7-epi-aciphyllene (123) obtained from (R)-limonene ([α]D27=+13.5, c 1.3, CHCl3),318 and (−)-γ-gurjunene (ent-120) made accessible through an enantioselective Morita-Baylis–Hillman reaction using an enantiopure phosphine catalyst ([α]D20=−121.1, c 0.1 CHCl3).322 For both compounds full NMR data were provided.318, 322
6 Cyclised and Rearranged Guaianes
Further cyclisations eventually with skeletal rearrangements are important for two groups of compounds originating from J1 and J3, while no examples from the other cations of the J series or from cations of the K series are known.
6.1 Compounds from J1
Compounds from J1 include patchouli alcohol (129), the patchoulenes 130–133 and seychellenes 134 and 135 (Scheme 35 A). The common biosynthesis of these compounds can be understood from J1 by a long range proton shift from C1 into the isopropenyl group to J1 c, followed by cyclisation to J1 d (path a) and deprotonation to β-patchoulene (130) and δ-patchoulene (131). An alternative cyclisation from J1 c to J1 d (path b) and deprotonation yields α-patchoulene (132) and γ-patchoulene (133). A Wagner–Meerwein rearrangement of J1 e to J1 f gives access to patchouli alcohol (129) by attack of water, while a methyl group migration to J1 g and deprotonation results in seychellene (134) or cycloseychellene (135). This pathway is in agreement with feeding experiments using radioactively labelled (4R)-[2–14C,4-3H]mevalonic acid,323, 324 and with deuterium incorporation from (2-2H)FPP at C5 of 129 and several side products from patchoulol synthase,22, 290 while a reported additional deuteration at C15 is difficult to understand.

A) Biosynthesis of cyclised and rearranged guaianes from J1. B) Initially assigned structures for (−)-patchouli alcohol (129 a and 129 b) and cycloseychellene (135 a).
Patchouli alcohol or patchoulol (−)-129 was first isolated as the main constituent from patchouli oil (P. cablin) in 1869.325 The oil is one of the most important industrial fragrances that is widely used in perfumery and cosmetics products. Its planar structure was described more than 80 years later as that of 129 a (Scheme 35 B).326 A structural revision based on chemical transformations and a synthesis from (+)-camphor through 132 resulted in the assignment of structure 129 b.327-329 However, a subsequent X-ray analysis of the chromic acid diester surprisingly led to the structure of 129,330 suggesting that during the synthesis of this compound from 132 a similar skeletal rearrangement as in the biosynthesis must have taken place. A later synthesis from (R)-carvone (ent-38) resulted in (−)-129 ([α]D25=−121.3, c 2.3, CHCl3).331 Compound (−)-129 was also isolated from plants of the genera Valeriana332-334 and Nardostachys335, 336 The complete 13C NMR data of 129 are available.290, 333, 337
The patchoulenes 130–133 and seychellenes 134 and 135 have been reported to co-occur with 129 in several species,307, 332, 334-336, 338, 339 and also many of these compounds are observed as products of the patchoulol synthase,22, 290 supporting their common biosynthesis through shared intermediates (Scheme 35 A) and corresponding absolute configurations. Formally, the absolute configuration of 130 ([α]D30=−42.6, c 10.51, CHCl3) was specified by chemical correlation with patchouli alcohol through acid treatment, at a time when 129 b was believed to be the correct structure of this sesquiterpene alcohol. Pyrolysis of patchoulyl acetate (135) yielded a mixture of 132 and 133, and dehydration with POCl3 resulted in a mixture of mainly 132 with 130 and 133.328 A reinterpretation of the results from these experiments included a Wagner–Meerwein rearrangement (Scheme 36).340 Compound 131 was first obtained by the acid-catalysed transformation of 129341 and later isolated from patchouli oil.342 The complete 1H and 13C NMR data of 130 are available,308 while those of 131–133 are lacking.

A) Acid promoted conversion of 129 into 130. B) Pyrolysis of patchoulyl acetate (135) to patchoulenes 132 and 133.
Seychellene (134, Scheme 35 A), [α]D=−72 (c 0.4, CHCl3),343 was first found in patchouli oil (“hydrocarbon G”),307 followed by structure elucidation through chemical degradation.340, 343 A total synthesis of (−)-134 from (R)-carvone (ent-81) confirmed its absolute configuration.344 Cycloseychellene (135) was reported to possess the structure of 135 a (Scheme 35 B) when it was first isolated from P. cablin in 1973.339 In 1981, Welch et al. synthesised (±)-135 a and found that the spectral and chromatographic properties of the synthetic hydrocarbon differed significantly from those of the natural product.345 A re-examination of the NMR spectra of cycloseychellene indicating that its structure should be corrected to that of 135.346 The 1H- and 13C-NMR data of 134 are available from the literature.308, 344
6.2 Compounds from J3
The biosynthesis of rotundene (136), isorotundene (137) and cyperene (138) can be understood from J3 (Scheme 37 A). Its cyclisation to J3 a (path a) and deprotonation yields 136 and 137, while a 1,2-hydride shift to J3 b (path b) followed by a 1,5-proton shift to J3 c, cyclisation to J3 d and deprotonation result in 138. This common biosynthetic pathway nicely explains the co-occurrence of 136–138 in Cyperus rotundus.347 Compound 136 ([α]D=−16.3) was first reported from C. rotundus and C. scariosus,348 and later also from C. alopecuroides,349 but at this stage only with the planar structure. (−)-Isorotundene (137) was isolated from C. rotundus whose relative configuration was determined by NOESY.347 This allowed to demonstrate that 136 has the same skeleton by conversion into rotundol (139) through oxymercuration and dehydration with POCl3 (Scheme 37 B). The absolute configuration of 136, and thus also of 137, was determined by ozonolysis to 140, decarboxylation to a mixture of epimers 141 ab, Wittig methylenation to 142 ab and catalytic hydrogenation to 119 ab (Scheme 37 C). One of these hydrocarbons was identical to 119 a obtained by hydrogenation of 120 (Scheme 30 C). Complete 1H- and 13C-NMR data for 137 have been reported,347 but are lacking for 136.

A) Biosynthesis of 136–138 from J3. Chemical correlation of B) 136 to 137, and C) 136 to 119 a, the hydrogenation product of (+)-γ-gurjunene.
The sesquiterpene 138 ([α]D20=−20.0, neat), was first isolated from Cyperus rotundus.350, 351 Its absolute configuration was resolved by the chemical correlation through its hydrogenation product that was identical to a material derived from 129 by dehydration with POCl3 and hydrogenation.352, 353 The (−)-enantiomer of 138 was later isolated from several other plants.177, 349, 354-367 Full 1H- and 13C-NMR data in CDCl3 and C6D6 have been reported.367, 368
7 Conclusions
Germacrene A shows a unique and interesting chemistry mainly characterised by its reactivity towards acid-catalysed cyclisations and its thermal lability in a Cope rearrangement to β-elemene. Similar observations have been made for other germacrenes,369 suggesting that the high ring strain associated with the 10-membered ring in these systems may be a strong driving force for the observed reactions leading to much less strained compounds with 6-membered rings. The reactivity built up by the ring strain is also used in enzymatic reactions towards sesquiterpenes for which germacrene A serves as an important intermediate. In enzyme reactions not only the formation of 6–6 bicyclic compounds, but also of 5–7 bicyclic derivatives can be achieved, and for both cases follow-up chemistry by skeletal rearrangements can further increase the structural variability. Subsequent steps include oxidative and other modifications after terpene cyclisation, leading to numerous derivatives for each compound presented in this review, which further underlines the central importance of germacrene A in sesquiterpene biosynthesis.
Acknowledgements
Open access funding enabled and organized by Projekt DEAL.
Conflict of interest
The authors declare no conflict of interest.