Stereoselective synthesis of trans-fused iridoid lactones and their identification in the parasitoid wasp Alloxysta victrix, Part I: Dihydronepetalactones

Starting from the enantiomers of limonene, all eight stereoisomers of trans-fused dihydronepetalactones were synthesized. Key compounds were pure stereoisomers of 1-acetoxymethyl-2-methyl-5-(2-hydroxy-1-methylethyl)-1-cyclopentene. The stereogenic center of limonene was retained at position 4a of the target compounds and used to stereoselectively control the introduction of the other chiral centers during the synthesis. Basically, this approach could also be used for the synthesis of enantiomerically pure trans-fused iridomyrmecins. Using synthetic reference samples, the combination of enantioselective gas chromatography and mass spectrometry revealed that volatiles released by the endohyperparasitoid wasp Alloxysta victrix contain the enantiomerically pure trans-fused (4R,4aR,7R,7aS)-dihydronepetalactone as a minor component, showing an unusual (R)-configured stereogenic center at position 7.


Introduction
The endohyperparasitoid wasp Alloxysta victrix is part of the tetratrophic system of oat plants (Avena sativa), grain aphids (Sitobion avenae), primary parasitoids (Aphidius uzbekistanicus) and hyperparasitoids (Alloxysta victrix). Chemical communication by volatile signals is considered to play a major role in interactions between these trophic levels, and some semio-chemicals of the lower trophic levels such as oat plant and grain aphid have been identified [2][3][4]. However, there is nearly no information about the intra-and interspecific signaling pathways between primary parasitoids and hyperparasitoids. In order to gain further information about the chemical structures and the biological significance of corresponding signals, we examined the volatile components of pentane extracts from dissected heads as well as headspace volatiles of Alloxysta victrix by coupled gas chromatography/mass spectrometry (GC/MS). Figure 1 shows one major component and several trace components which could be identified as 6-methyl-5hepten-2-one (1), neral (2), geranial (3), actinidine (4), and geranylacetone (5). Bioassays revealed the main compound 1 to be repellent to the aphid-parasitoid, Aphidius, by warning the primary parasitoid of the presence of the hyperparasitoid [5]. The prenyl-homologue of 1, geranylacetone (5), seems to be a component of the sex pheromone of Alloxysta victrix [6]. In addition, GC/MS of cephalic secretions of both sexes showed a minor component X, the mass spectrum of which suggested it to be a trans-fused dihydronepetalactone. Since no synthetic reference compounds were available, we had to synthesize all eight trans-fused dihydronepetalactones to unambiguously identify the natural product X. The realization of this task is the subject of the present paper.

Results and Discussion
Apart from a couple of known acyclic terpenoids (Figure 1), analysis by gas chromatography coupled with mass spectrometry (GC/MS) revealed the presence of an unknown minor component X in both sexes of Alloxysta victrix. Chemical ionization analysis (GC/CIMS) showed the molecular mass of the compound to be M + = 168, while high resolution mass spectrometry (GC/HRMS) proved its atomic composition to be C 10 H 16 O 2 , suggesting an oxygenated monoterpene as the target structure. The fragmentation pattern, exhibited in the 70 eV EI-mass spectrum (Figure 2), showed some similarities to that of the known cis-fused dihydronepetalactone (6) [7], however, differences in relative abundances of fragment ions pointed to a trans-fused dihydronepetalactone as the target structure [8]. In the mass spectrum of the cis-fused compound m/z 67 and m/z 95 were of similarly low intensity (30%), while in that of the unknown natural product X the two fragments were highly abundant (80%). The most striking differences in the spectra were pronounced signals for the molecular ion M + = 168 and M + − 15 (70% and 40%, respectively) for the cis-fused dihydronepetalactone whilst in the spectrum of X the two signals were of only low abundance ( Figure 2). Dihydronepetalactones are derivatives of nepetalactone (7) which was first isolated by Mc Elvain in 1941 from the essential oil of catmint, Nepeta cataria ( Figure 1) [9]. Relative configurations of cis-fused nepetalactones and some related derivatives have been investigated [10,11]. Nepetalactone and cisas well as trans-fused dihydronepetalactones have been isolated from the leaves and galls of the plant Actinidia polygama [8]. In addition, dihydronepetalactones are components of the defensive secretions of some ant species [12], while nepetalactone and the corresponding lactol showing (1R)-configuration have been identified as pheromones of aphids [13,14]. (1R,4S,4aR,7S,7aR)-Dihydronepetalactol (8) was characterized as a semiochemical for lacewings [15].
The dihydronepetalactone skeleton shows four contiguous stereogenic centers, giving rise to eight trans-fused stereoisomers a-d and the corresponding enantiomers a'-d' (Figure 3).
Route to trans-fused dihydronepetalactones a and b starting from (S)-pulegone For comparison, the synthesis of a and b was carried out following Wolinsky's approach: (S)-Pulegone (9) was transformed to trans-pulegenic acid 10 via bromination, Favorskii rearrangement, and subsequent elimination (Scheme 1). Stereo-selective addition of hydrochloric acid afforded the chloride 11, and subsequent elimination of hydrochloric acid gave a mixture of the methyl esters 12 and 13 (methyl trans-pulegenate) [20][21][22] which could be separated by chromatography on silica gel. Hydroboration and lactonization of 12 furnished a mixture of the C4-epimers a and b that once again needed to be separated by chromatography on silica gel [23].
Analytical data of the first eluting component a were in accordance with those reported in the literature [24]. The same sequence starting from (R)-pulegone yielded a mixture of diastereomers a' and b'. The relative configuration of a at C4 was assigned according to NOESY experiments. Decisive NO-effects were found between the protons 4-H and 7a-H as well as between 7a-H and 7-CH 3 ( Figure 4). Basically, the sequence developed by Wolinsky could also provide access to the diastereomers c and d (and their enantiomers) if trans-pulegenic acid (10) would be replaced by cis- pulegenic acid. A mixture of the latter and its trans-isomer (60:40) can be obtained by using a different base in the Favorskii-rearrangement step [22], again requiring a difficult chromatographic separation. Furthermore, this multistep route has several major disadvantages: The formation of mixtures of epimers entails to separations at several stages which have proven to be problematic. Moreover, several reaction steps afford unsatisfactory yields [23]. In addition, one of the main disadvantages is the fact that (S)-pulegone (S-9) is a highly expensive starting material for the synthesis of four of the eight trans-fused dihydronepetalactones. That excludes this route for the synthesis of larger amounts.

Route to stereochemically pure trans-fused dihydronepetalactones from (R)-limonene
Due to the shortcomings of the route described above, we designed an improved strategy towards trans-fused dihydronepetalactones. Starting from 1-formyl-2-methyl-5-(1methylethenyl)-1-cyclopentene (15) as the key intermediate, the stereoselective synthesis of all eight stereoisomers could be achieved ( Figure 5). The aldehyde 15 could be readily prepared from commercially available pure and cheap (R)-limonene (14) [25][26][27]. Non-selective hydroboration of the double bond in the side chain of 15 would yield a pair of diastereomers 16/16* which would have to be separated. However, we expected that the chiral center at C5 would cause stereocontrol by forcing the reaction to proceed through the sterically least hindered tran-sition state. We envisioned that stereoselective hydrogenation of the endocyclic double bond of the key intermediate 16 (and its diastereomer 16*) in either a "syn"-or "anti"-fashion could yield two pairs of diastereomeric hydroxy carboxylic acids 17/17* or 18/18* after some simple functional group transformations. These hydroxy acids would then yield the desired trans-fused dihydronepetalactones a-d during a final lactonization step.
Starting from cheap and pure (S)-limonene (14'), the corresponding trans-fused dihydronepetalactones a'-d' could be synthesized in the same way, showing our novel route to be a versatile alternative to Wolinsky's sequence [20,21]. In contrast to the latter, which fixed the stereogenic center of pulegone at position 7 of the final dihydronepetalactone, in our route the stereogenic center of limonene is retained at position 4a of the target compound and used for the stereoselective introduction of additional chiral centers.

Synthesis of trans,trans-dihydronepetalactone b
To install a trans,trans-configuration between the substituents at C5-C1 and C1-C2 of the cyclopentane backbone -which would later reflect the trans,trans relationship between substituents at C7-C7a and at C7a-C4a of the dihydronepetalactones a and b -a formal "anti"-addition of hydrogen to the cyclopentene 16 had to be carried out (Scheme 3). Usually, both homogeneous and heterogeneous catalytic hydrogenation reactions proceed via "syn"-addition of hydrogen to olefinic double bonds. Only subsequent isomerization processes may lead to a formal "anti"-addition. To obtain a suitable precursor, which, due to enolization of the hydrogenation product, might allow this formal "anti"-addition of hydrogen, the key intermediate 16 was transformed to the aldehyde 23.
In the course of this short sequence, the free hydroxy group of 16 was protected as the TBDMS ether to yield 21 which afforded the mono-protected diol 22 after treatment with KOH in methanol. Subsequently, 22 was oxidized with pyridinium dichromate to give aldehyde 23.
We expected that catalytic hydrogenation of the trisubstituted cylopentene 23 with a heterogeneous catalyst would preferentially take place from the sterically less hindered side of the molecule. This would lead to an all-cis configured hydrogenation product which would endure considerable steric strain. Due to the CH-acidity at the α-position of the formyl group, epimerization of the all-trans product 24 under acidic or basic conditions could be expected. Lange et al. reported the catalytic hydrogenation of a structurally close analogue, (5R)-1-formyl-2-methyl-5-isopropylcyclopent-1-ene, over Pd/C (10%) to give a 9:1 mixture of the all-cis versus the all-trans product [29]. In our case, the application of Lange's method to the aldehyde 23 led to the formation of a 3:1 mixture of the all-cis versus the alltrans epimer. Subsequent treatment with sodium methoxide in MeOH at rt for 20 h completely shifted the equilibrium to the thermodynamically more stable all-trans product 24. Unfortunately, these results could not be reproduced on larger reaction scales (>5 mmol). After screening of a variety of other hydrogenation conditions, we found the hydrogenation of 23 with ammonium formate over palladium on carbon (10%) to be the method of choice [30]. Using this approach, the all-trans aldehyde 24 was almost exclusively formed. The presence of ammonium formate in the reaction mixture probably leads to an "in-situ" epimerization at C2 from the kinetically formed all-cis to the thermodynamically more stable all-trans product. Relative configurations of all substituents of compound 24 were confirmed by NOESY experiments (Figure 6). Strong NO-effects were found between the protons 5-CH 3 and 1-H, 1-H and 1'H (the proton at C1 of the side chain), 5-H and 2-H as well as between protons of 1-CHO and 2-H which is in line with a trans,trans-configuration (using the nomenclature described above).
Starting from 24, the trans,trans-dihydronepetalactone b was synthesized in three subsequent steps (Scheme 3). First, oxidation of the aldehyde group with potassium permanganate in the presence of a phosphate buffer (pH 4.5) afforded the carboxylic acid 25 without epimerization at C1 [23,31]. Subsequent deprotection of the TBDMS ether with tetrabutylammonium fluoride (TBAF) yielded 17, and lactonization with N,Ndicyclohexylcarbodiimide (DCC) and 4-dimethylaminopyridine (DMAP) in dichloromethane afforded the trans,transdihydronepetalactone b.
The relative configuration of the trans,trans-dihydronepetalactone b was confirmed by NOESY experiments. Decisive NO-effects could be observed between 4-CH 3 and 7a-H as well as between 4-CH 3 and 5-Ha, and furthermore, between 5-Ha and 7a-H as well as between 7a-H and 7-CH 3 ( Figure 6). The enantiomer b' was synthesized via the same route starting from (S)-limonene. Analytical data of compound b' were identical to those which were obtained of b when Wolinky's route was followed (see above).

Synthesis of cis,trans-fused dihydronepetalactone c
For the synthesis of the cis,trans dihydronepetalactone c, a cis,trans-configuration between the substituents at C5-C1 and C1-C2 of the cyclopentane backbone needed to be established. With acetate 16 as the key intermediate, a stereoselective "syn"addition of hydrogen from the same side as the (R)-configured side chain at C5 would provide the desired stereochemical outcome of the hydrogenation reaction (Scheme 4). We expected the free hydroxy group of 16 to coordinate to an appropriate homogenous hydrogenation catalyst, controlling the stereochemical course of the hydrogen transfer from the same side as the side chain at C5 through chelation. We chose Crabtree's iridium catalyst ([Ir(cod)PCy 3 (py)]PF 6 ) which has been reported to furnish excellent facial selectivities during directed hydrogenations of cyclic olefins [32][33][34].
Hydrogenation of acetate 16 in the presence of 11 mol % of Crabtree's catalyst under 1 bar hydrogen pressure for 1.5 h yielded the desired product 26 as a single diastereomer. Alternative hydrogenation methods using optically active catalysts failed. In one case we investigated the hydrogenation of the endocylic double bond of the allylic alcohol 19 (Scheme 2) with one of Noyori's ruthenium BINAP catalysts ([Ru((S)-BINAP)](OAc) 2 ) [35,36] but reduction occurred only at the side chain.
Relative configurations of all substituents of the acetate 26 were confirmed by NOESY experiments (Figure 7). Strong NO-effects were observed between the protons 5-CH 3 and 1'-H (protons of the acetoxymethyl group at C1), 5-CH 3 and 2-H, 1'-H and 2-H as well as 5-H and 1''-CH 3 (protons of the methyl group at C1 of the side chain) which is in line with a cis,transconfiguration (using the nomenclature described above).
Starting from 26, the synthesis of the cis,trans-dihydronepetalactone c was completed in five subsequent steps (Scheme 4). First, the free hydroxy group of 26 was protected as the TBDMS ether to yield 27. Then, the acetate group was removed with methanolic KOH to afford the alcohol 28. Careful oxidation of the primary alcohol function [37,38] with ruthenium(III) chloride and sodium periodate in a biphasic mixture of carbon tetrachloride, acetonitrile, and phosphate buffer (pH 7) produced the carboxylic acid 29 without epimerization at C1. After removal of the TBDMS protecting group with HF in acetonitrile, the hydroxy acid 18 was lactonized in the presence of DCC and catalytic amounts of DMAP in dichloromethane at rt to afford cis,trans-dihydronepetalactone c. Its enantiomer c' was synthesized from enantiomerically pure (S)-limonene, following the same route. The relative configuration of c was confirmed by NOESY experiments (Figure 7). Decisive NO-effects could be observed between 4a-H and 7-CH 3 as well as between 4-CH 3 and 7a-H, and furthermore, between 4-CH 3 and 5-Ha as well as between 5-Ha and 7a-H.

Synthesis of a mixture of trans-fused dihydronepetalactones c and d
The stereogenic center at C1' of the acetate 26 keeps (R)-configuration which resulted from highly stereoselective hydroboration of the acetate 20 to yield the key intermediate 16 as shown above (Scheme 2). For the synthesis of the cis,transdihydronepetalactone d, this stereocenter needed to be isomerized to keep the (4S)-configuration in the final product (Scheme 5). To achieve the required inversion, the acetate 26 was oxidized to the aldehyde 30, which could be epimerized using p-toluenesulfonic acid in benzene under reflux conditions to provide a 2:3 mixture of the desired aldehydes 30 and its epimer 30*. Subsequent steps were carried out with the mixture of diastereomers. Reaction of 30/30* with sodium borohydride at −20 °C reduced the aldehyde function to yield a mixture of the diastereomers 26/26*. The following sequence, yielding a mixture of the dihydronepetalactones c and d was essentially the same as described above (Scheme 4).
Transformation of the free hydroxy group to the TBDMS ethers 27/27* was followed by cleavage of the acetate moiety with methanolic KOH to give a mixture of the alcohols 28/28*. Oxi- dation of the primary alcohol function with ruthenium(III) chloride and sodium periodate in a biphasic mixture of carbon tetrachloride, acetonitrile and phosphate buffer (pH 7) afforded the carboxylic acids 29/29* without epimerization at C1 [37,38]. After cleavage of the TBDMS ether with HF in acetonitrile, a mixture of dihydronepetalactones c and d was formed by lactonization of the hydroxy acids 18/18* with DCC and DMAP in dichloromethane at rt. The C4 epimeric dihydronepetalactones c and d could be separated by column chromatography over silica. Starting from the enantiomer of 26, a mixture of dihydronepetalactones c' and d' was synthesized by following the same reaction sequence.
Formal synthesis of a mixture of trans-fused dihydronepetalactones a and b from (R)-limonene As outlined above, six of the eight possible stereoisomers of trans-fused dihydronepetalatones were synthesized from the enantiomers of limonene following a new route. Compound a and its enantiomer a' were prepared according to the procedure described by Wolinsky [20,21]. However, our new approach also includes a formal synthesis of a and a'. A mixture of a and b will be easily obtained from the protected hydroxy aldehyde 24 by the straight forward procedure outlined in Scheme 6.
Reduction of the aldehyde function of 24 and acetylation of the resulting primary alcohol followed by cleavage of the silyl group will furnish the primary alcohol 31, which upon oxidation will yield the corresponding aldehyde that can be epimerized to the diastereomers 32/32* as shown above. Subsequent reduction of 32/32*, silylation of the resulting primary alcohols and saponification will produce a mixture of the diastereoisomers 33/33*. Oxidation of the primary alcohol moiety, followed by cleavage of the silyl group will yield the epimeric hydroxy acids 34/34* which will form a mixture of the dihydronepetalactones a and b after lactonization. As shown above, this mixture can be separated upon column chromatography.

Conclusion
In summary, we synthesized all eight trans-fused stereoisomeric dihydronepetalactones. After having used the enantiomers of pulegone as educts in Wolinsky's route to (4S,4aS,7S,7aR)-dihydronepetalactone (a) and its enantiomer a' [23], we developed an improved and general way for the synthesis of all trans-fused dihydronepetalactones, starting from pure enantiomers of limonene. Our approach is also superior to that starting from optically active carvone that yields the starting material for the synthesis of trans-fused iridoid lactones only as a byproduct [15].

Identification of a trans-fused dihydronepetalactone in the parasitoid wasp Alloxysta victrix
Upon gas chromatography using FFAP as a polar achiral stationary phase, the stereoisomers a and c could be well separated while b and d coeluted. However, the latter pair could be resolved on a less polar DB5-capillary, where b/b' eluted after d/d' (data not shown). As a result, the relative configuration of each of the trans-fused dihydronepetalactones could be unambiguously assigned by GC/MS.
With the exception of (4S,4aS,7R,7aR)-dihydronepetalactone (d) and its enantiomer d', the stereoisomers could well be distinguished by enantioselective gas chromatography using a 1:1-mixture of OV1701 and heptakis-(6-O-tert-butyldimethylsilyl-2,3-di-O-methyl)-β-cyclodextrin as an optically active stationary phase. Figure 3 shows the corresponding retention times of all eight stereoisomers that were obtained with the two used capillary column systems. Coupled GC/MS using FFAP as the stationary phase revealed the target natural iridoid lactone X to show the same mass spectrum and the same retention time as a/a', the first eluting pair of the synthetic dihydronepetalactones ( Figure 3). Enantioselective gas chromatography on a cyclodextrin column showed X to coelute with a' which was well separated from its enantiomer by an α-value of a':a = 1.01 ( Figure 3). Consequently, the structure of X was unambiguously assigned to be (4R,4aR,7R,7aS)-dihydronepetalactone. It should be noted that Meinwald et al. identified a/a' (absolute configuration not assigned) as a component of secretions of the abdominal defense glands of the rove beetle Creophilus maxillosus [39]. Interestingly, the structure of a' is relatively close to that of nepetalactone 7 and lactol 8, the sex pheromone of the grain aphid S. avenae [13] which keeps the second level in the investigated tetratrophic system. Grant et al., found the trans-fused (1R,4aS,7R,7aR)-1-methoxy-4,7dimethyl-1,4a,5,6,7,7a)-hexahydrocyclopenta[c]pyran, called (1R)-1-methoxymyodesert-3-ene, among the volatiles of the Ellangowan poison bush, which they transformed to the corresponding lactone a' [24]. Apart from this compound and very few others, the stereogenic center carrying the methyl group in the five-membered ring of iridoid lactones including insect semiochemicals [13][14][15] generally shows (S)-configuration. Only recently, two isomeric iridoid lactones showing (7R)-configuration have been identified from the Drosophila parasitoid Leptopilina heterotoma [40]. Compound X has been identified in the mandibular gland secretions of other Alloxysta species, too, [41]. However, its biological significance is not yet clear and will need further investigations.
The differentiation of the oxygen containing functional groups in the trisubstituted cyclopentene 16, a key-compound in our synthetic approach, provides access to a large number of iridoids including nepetalactones but also iridomyrmecins and monocyclic compounds. Consequently, having reference compounds at hand, structures of hitherto unknown iridoids [42] may now be assigned. It may turn out that the chiral center carrying the methyl group in the five-membered ring of iridoids may much more often show (R)-configuration than it is known today.

Supporting Information
Supporting Information File 1 Experimental details and characterization data for synthesized compounds.