Total Synthesis of Avenaol

Abstract

Avenaol is a terpene with a unique all-cis cyclopropane in which all bulky substituents are oriented in the same direction. It is categorized into a non-canonical strigolactone. We have synthesized alkylidenecyclopropanes by Rh-catalyzed intramolecular cyclopropanation of allenes, followed by iridium-catalyzed diastereoselective double bond isomerization to construct all-cis cyclopropanes. Subsequently, distinction of the two hydroxymethyl groups of 1,3-diol by an intramolecular S_N1-type reaction, followed by cleavage of the tetrahydropyranyl ring by regioselective C–H oxidation, led to the desired stereochemistry at the C-ring lactone. Using these key steps, the first racemic total synthesis of avenaol was achieved, and the proposed relative configuration of avenaol was proved synthetically. Furthermore, we developed a stereoselective introduction of D-ring butenolide via chiral thiourea-quaternary ammonium salt-catalyzed dynamic kinetic optical resolution. Then, by applying this method to synthetic intermediates, (+)-avenaol was successfully synthesized. This chapter details the total synthesis of avenaol, including failed attempts.

18.1 Introduction

18.1.1 Structure and Properties of Avenaol

Strigolactones (SLs) are plant-produced terpenes that act as rhizosphere-signaling substances to induce mycelial branching in arbuscular mycorrhizal fungi and are responsible for the symbiotic relationship between plants and arbuscular mycorrhizal fungi [1]. They also inhibit branching and are recognized as plant hormones. The basic skeleton of a SL is composed of a tricyclic lactone (ABC ring moiety) and a butenolide (D ring) with an acetal, which are connected via an enol ether, as represented by strigol and orobanchol (Fig. 18.1a). In 2014, avenaol (1), a SL with a novel structure that differs from the basic skeleton, was isolated from root secretions of Avena strigosa by Yoneyama et al. (Fig. 18.1b) [2]. The structural features of avenaol include (1) the AB-fused ring system is a bicyclo[4.1.0]heptanone skeleton in which 3- and 6-membered rings are fused, (2) the B ring is a cyclopropane with all bulkier substituents oriented in the same direction (this structure is hereinafter labeled as all-cis cyclopropane), (3) the number of carbons is one more than other SLs with a typical parent skeleton, and (4) the enol ether structure connecting the C and D rings is common to other SLs. The relative stereochemistry has been determined by 2D NMR, including COSY and NOESY, and the absolute stereochemistry has been inferred from homology with related SLs. Avenaol shows seed germination-stimulating activity. Although seeds of Orobanche minor and Striga hermonthica did not germinate in a 10 nM solution, 49% germination was observed for seeds of Pinguicula ramose. We have conducted synthetic studies of avenaol to confirm its unique structure and its relative and absolute stereochemistry, to elucidate the structure–activity relationship using synthetic analogs, and to develop a new method for the construction of the all-cis cyclopropane ring.

Fig. 18.1

Structures of SLs and avenaol

18.1.2 Preliminary Investigations

The ABC ring skeleton of avenaol is completely different from that in other natural products, while the enol ether structure is common to other SLs. Therefore, the challenges for the synthesis of avenaol are (1) the construction of a bicyclo[4.1.0]heptanone skeleton containing an all-cis cyclopropane (i.e., the AB ring skeleton) and (2) the control of stereochemistry at C8 of the C-ring moiety and C3 of the A-ring moiety [3,4,5,6,7,8,9,10,11,12,13,14]. Initially, cyclopropanation by the Corey–Chaykovsky reaction was attempted for construction of the AB ring skeleton [3]. Cyclohexenone 2 and sulfonium salt 3 were treated with various bases in several solvents, but the desired product with a bicyclic skeleton was not obtained (Scheme 18.1a) [15]. For the synthesis of the bicyclo[4.1.0]heptanone skeleton, intramolecular cyclopropanation of diazo compounds had been reported to give a product with a cage-like structure. Thus, we attempted intramolecular cyclopropanation of diazo compounds with trisubstituted alkenes. The reaction of cyclic alkene 5 with Rh₂(cap)₄ yielded only a complex mixture and no bicyclic product (Scheme 18.1b), and the reaction with Cu(tbs)₂ did not yield the desired product and only dimerized product 8 was observed (Scheme 18.1b). Other non-cyclic alkenes 9 were also examined, but the cyclopropanation did not proceed (Scheme 18.1c). Comparing these results with previous reports by Corey et al., the positions of the substituents on the olefins are different [4]. Thus, it was assumed that cyclopropanation of 5 and 9 did not proceed because of steric repulsion between the substituent methyl group and the metal carbenoid in the transition state, which would make the cyclopropanation pathway unfavorable.

Scheme 18.1

Preliminary investigation of cyclopropanation of trisubstituted cyclopropanes

These initial investigations suggested that the construction of the bicyclo[4.1.0]heptanone skeleton by cyclopropanation of these alkenes was not suitable for avenaol synthesis.

18.2 Racemic Total Synthesis of Avenaol

18.2.1 Construction of Alkylidenecyclopropane

Because of the difficulties encountered with cyclopropanation of trisubstituted alkenes, other approaches were required to synthesize the bicyclo[4.1.0]heptanone skeleton with an all-cis structure. It was important to set appropriate synthetic intermediates taking into consideration the risk for cleavage of the cyclopropanes by introducing electron-withdrawing groups and the suppression of the formation of cage-like structures [12, 16]. Therefore, we focused on an alkylidenecyclopropane as a key intermediate to synthesize all-cis cyclopropanes. In a previous report, Sarpong et al. successfully constructed the bicyclo[3.1.0]hexanone skeleton by intramolecular cyclopropanation of an allene [17]. Charette et al. showed that the reactivities of metal carbenes in intermolecular cyclopropanation depend on the substituent on the carbene carbon [18]. According to these reports, we expected to synthesize alkylidenecyclopropanes by an intramolecular cyclopropanation of diazo compounds with allenes, following a diastereoselective conversion to all-cis cyclopropanes.

The retrosynthetic analysis is shown in Scheme 18.2. Avenaol (1) would be constructed by coupling bromobutenolide 11a, which is the D-ring moiety, with enol 12, as in other SL syntheses [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35]. Enol 12 would be synthesized from lactone 13 via the stereoselective introduction of a hydroxy group on C3. Lactone 13 would be accessed by constructing the C ring and introducing two carbon units into 14. The all-cis cyclopropane of 14 would be constructed from alkylidenecyclopropane 15. Although various approaches might be possible to convert alkylidenecyclopropanes to the all-cis structure, we initially planned to use hydrogenation because it is the simplest method. As described above, alkylidenecyclopropanes 15 would be synthesized by intramolecular cyclopropanation of diazoketone 16 with dimethyl groups and oxygen functionality. Diazoketone 16 would be accessed from a known aldehyde 17, which could be prepared in one step from inexpensive starting materials.

Scheme 18.2

Retrosynthesis through hydrogenation of alkylidenecyclopropane

To investigate intramolecular cyclopropanation, diazoketone derivatives 16a–e with a methyl group, a nitrile, and an ester as a substituent were synthesized (Scheme 18.3). The known aldehyde 17 was treated with tetrahydropyranyl (THP)-protected propargylic alcohol and benzyltrimethylammonium hydroxide to afford the secondary alcohol 18a [36, 37]. This transformation could be replaced with asymmetric nucleophilic addition. Following Carreira’s procedure, treatment of 17 with p-methoxybenzyl (PMB)-protected propargylic alcohol in the presence of Zn(OTf)₂ and (−)-methylephedrine as a ligand successfully gave alcohol 18b [38]. Racemic 18a was used for further investigation because of ease and cost of the synthesis. The secondary hydroxy group of the resulting 18a was converted to the propargylic alcohol 19 by methylation and acidic treatment to remove the THP group. Hydroalumination of 19 followed by treatment with iodine gave allene 20 [39]. After the formation of a benzyl ether from 20, selective hydroboration of a terminal olefin, followed by oxidation to carboxylic acid 21a by reaction with either 9-azanoradamantane N-oxyl (nor-AZADO) or sulfur trioxide pyridine complex followed by NaClO₂ [40]. Similarly, methoxymethyl (MOM) ether 21b and triisopropylsilyl (TIPS) ether 21c were synthesized. Diazo ketone 16a was obtained in low yield via conversion of 21a to acid chloride with Ghosez reagent, followed by treatment with freshly prepared diazoethane and 4-dimethylaminopyridine (DMAP). Carboxylic acid 21a was converted to a β-ketoester using Masamune’s procedure and then to β-keto-α-diazo ester 16b by diazotransfer with 4-acetamidobenzenesulfonyl azide (ABSA) [41, 42]. After esterification of 21a–c via acid anhydrides and conversion to β-ketonitriles, β-keto-α-diazonitriles 16c–e were synthesized by diazotransfer using (imid)SO₂N₃ [43].

Scheme 18.3

Synthesis of diazo compounds bearing an allenyl group

Next, the rhodium- and copper-catalyzed intramolecular cyclopropanation of allenes 16a–e was investigated to synthesize alkylidenecyclopropanes 15a–e (Table 18.1). Initially, methyl diazoketone 16a was used because 15a would not require subsequent functional group transformation, which would reduce the number of synthetic steps. However, treatment of 16a with a catalytic amount of Rh₂(OAc)₄ in dichloromethane gave no desired product and carboxylic acid 21a in 26% yield (entry 1). This result could be attributed to the instability of rhodium carbenes arising from 16a. Thus, β-keto-α-diazo ester 16b was used because a carbenoid derived from 16b was expected to be a more stable but still reactive. The reaction of 16b with Rh₂(OAc)₄ was unsuccessful (entry 2). In the case of Cu(CH₃CN)₄PF₆, the reaction gave a complex mixture (entry 3). On the other hand, when the substrate was changed to β-keto-α-diazonitrile 16c, the reaction smoothly proceeded to give the desired alkylidenecyclopropane 15c in 85% yield as a single diastereomer with a double bond in the E configuration (entry 4). The relative stereochemistry of 15c was determined by NOE correlation. Like Charette et al., we speculated that the high electrophilicity of the cyanorhodium carbene contributed to the acceleration of this reaction [18]. The cyclopropanation of 16d bearing MOM ether gave 15d in 96% yield, and the cyclopropanation of 16e bearing TIPS ether gave 15e in 99% yield (entries 5 and 6).

Table 18.1 Intramolecular cyclopropanation of allenes

Only (E)-olefins were obtained presumably because steric repulsion between the benzyl (or methoxymethyl or silyl) ether of the allene substituent and the ligand coordinating to the rhodium carbene favored the transition state TS-A over TS-B. Consequently, cyclopropanation would proceed from the opposite side of the benzyl ether (Scheme 18.4).

Scheme 18.4

Explanation for the diastereoselectivity

The cyano group of the resulting alkylidenecyclopropane 15d was converted to a methyl group. A ketone of 15d was diastereoselectively reduced to a secondary alcohol by treatment with sodium borohydride in the presence of cerium chloride, and then the alcohol was converted to PMB ether 22a (Scheme 18.5). The reason for the high diastereoselectivity in the first step is that the reductant approached from the opposite side of the face of the alkylidenecyclopropane. The cyano group of 22a was converted to an aldehyde by diisobutylaluminium hydride (DIBAL-H) reduction followed by treatment with sodium borohydride to give primary alcohol 23a. The Appel reaction of 23a gave an alkyl iodide which was treated with sodium borohydride at 80 °C in dimethyl sulfoxide (DMSO) to give a reduced product 24a [44]. Compound 24b was also synthesized from alkylidenecyclopropane 15e bearing a TIPS group by a similar route.

Scheme 18.5

Reduction of the cyano group to a methyl group

18.2.2 Initial Attempt to Construct an All-Cis Cyclopropane Using Hydrogenation and Radical Cyclization

Next, we investigated the construction of an all-cis cyclopropane from 24. Shibatomi and Iwasa et al. reported that a hydrogen source approached from the convex side of the oxabicyclo[3.1.0]hexanone skeleton in hydrogenation of alkylidenecyclopropanes [8]. Thus, we used these conditions to investigate whether hydrogenation of 24 would proceed in the same manner for the structurally similar bicyclo[4.1.0]heptanone skeleton (Scheme 18.6).

Scheme 18.6

Working hypothesis for hydrogenation

Unfortunately, hydrogenation of 24b in methanol yielded all-cis cyclopropane 14a and trans-cyclopropane 7-epi-14a as a 1:2 diastereomeric mixture (Scheme 18.7). After tetrapropylammonium perruthenate (TPAP) oxidation of this mixture, NOE correlations and coupling constants of the resulting ketone revealed that 7-epi-25 was the major product. The substrate used by Shibatomi and Iwasa et al. for the hydrogenation had no substituent in the angular position [8]. Thus, it was assumed that the methyl group in the angular position in 24b may have caused steric hindrance, which avoided the approach from the convex side.

Scheme 18.7

Hydrogenation of alkylidenecyclopropane

We also attempted radical reactions for the construction of all-cis cyclopropanes. When bromoacetal 27, which was derived from 24, is treated with a radical initiator and reductant, a 5-exo-trig radical cyclization reaction might occur [45]. If both the cyclization and reduction proceed from the convex side of the fused ring system, the stereochemistry on the C ring and the cyclopropane could be controlled in one step (Scheme 18.8).

Scheme 18.8

Retrosynthesis for the radical cyclization

After removal of the TIPS group of 24b, the resulting primary alcohol 24c was treated with ethyl vinyl ether and N-bromosuccinimide (NBS) to give bromoacetal 27 (Scheme 18.9). The obtained 27 was reacted with triethylborane as a radical initiator and tributyltin hydride as a reductant at − 78 °C under an oxygen atmosphere in toluene. This gave the cyclized product 26 in 63% yield as a diastereomeric mixture. The diastereomeric ratio was determined after derivatization to lactones 28 and 7-epi-28 via removal of the PMB group, hydrolysis of acetal, and TPAP oxidation. Unfortunately, the undesired trans-cyclopropanes 7-epi-28 were obtained as major products in diastereomeric mixtures on C8, along with small amount of all-cis cyclopropanes 28 (28:7-epi-28 = 1:11). The relative stereochemistry of the C7 of 7-epi-28 was determined from the coupling constant between C6 and C7. The coupling constant for a trans-cyclopropane was less than 6 Hz and different from the coupling constant (9.0 Hz) for avenaol (1) bearing all-cis cyclopropane.

Scheme 18.9

Attempt to construct all-cis cyclopropane by radical cyclization

The diastereoselectivity of this reaction could be rationalized as follows. The intramolecular addition of the alkyl radical to the olefin could proceed via conformation A or B (Scheme 18.10). Because the p-methoxybenzyloxy (PMBO) group on the cyclohexane ring of A and B was in an equatorial position, there was no steric hindrance around the double bond. Consequently, the addition proceeded through either A or B, and the selectivity on C8 was not expressed (diastereomeric ratio (dr) of 1.4:1 for C8). After formation of the five-membered ring, the reaction with the hydrogen source (M-H) would proceed via conformations C and D. Because these radicals in intermediates C and D have sp³ properties compared with olefins in A and B, steric repulsion would occur with the PMBO group in the equatorial position. Therefore, the reaction via D would be more favorable than the reaction via C.

Scheme 18.10

Explanation for the diastereoselectivity of radical cyclization

18.2.3 First-Generation Approach to All-Cis Cyclopropane Using Palladium-Catalyzed Reduction of Allyl Carbonates

Although hydrogenation and radical cyclization preferentially yielded undesired products with a trans-substituted cyclopropane, these results suggested that steric repulsion with substituents on the cyclohexane ring affected the selectivity. Therefore, we focused on the palladium-catalyzed reduction of allyl carbonates for constructing all-cis cyclopropane as follows [46, 47]. Lactone 13 would be synthesized by the diastereoselective 1,4-reduction of butenolide 29, which would be prepared from all-cis vinylcyclopropane 31 via dihydroxylation, esterification, and the intramolecular Horner–Wadsworth–Emmons reaction of phosphonate ester 30 (Scheme 18.11). It was envisioned that the all-cis cyclopropane structure of 31 would be constructed by diastereoselective reduction of allyl carbonate ester 24d because the bulky palladium center of a π-allyl palladium complex generated from 24d would be positioned at the outside of the molecule as shown in Scheme 18.11.

Scheme 18.11

Retrosynthesis for the Pd-catalyzed reduction of allyl carbonates

After conversion of allyl alcohol 24c to allyl carbonate 24d, construction of all-cis cyclopropanes was examined (Scheme 18.12). As expected, treatment of 24d with Pd(dba)₂, PBu₃, formic acid, and triethylamine in tetrahydrofuran (THF) under reflux gave all-cis cyclopropane 31 in 95% yield in a diastereoselective manner (9.5:1). Having obtained the all-cis cyclopropane with high diastereoselectivity, we then attempted to construct the C-ring moiety. Dihydroxylation of alkene 31 with OsO₄ was followed by condensation of the resulting diol 32 with diethyl carboxymethylphosphonate in a primary alcohol-selective manner. The secondary hydroxy group of the resulting phosphonate ester was converted to ketone 30 by TPAP oxidation, followed by treatment with potassium tert-butoxide in THF to afford butenolide 29a in 67% yield [74% based on recovered starting material (brsm)] via an intramolecular Horner–Wadsworth–Emmons reaction. The PMB group was removed by treatment with 2,3-dichloro-5,6-dicyano-p-benzoquinone (DDQ) to give the secondary alcohol 29b.

Scheme 18.12

Pd-catalyzed reduction of allyl carbonate and construction of the C ring

The 1,4-reduction of butenolides 29a and 29b was then examined. When 29b was treated with Crabtree’s catalyst in dichloromethane under a hydrogen atmosphere (6 atm), the reaction did not proceed (Table 18.2, entry 1). When magnesium metal was used as a single-electron reductant, a complex mixture was obtained (entry 2). On the other hand, when SmI₂ was used, trans-cyclopropane 7-epi-29b was obtained (entry 3) [48]. The hydride reductant was then investigated and the reaction with Stryker’s reagent did not proceed at all when the reaction was heated to reflux in benzene (entry 4) [49]. These results suggested that one-electron reductants and Cu–H are not effective for this transformation. On the other hand, the reaction with NaBH₄ in the presence of CoCl₂ gave the desired lactone 13b as a diastereomeric mixture, which was difficult to separate (entry 5) [50]. In the case of 29a with a PMB group, the reaction using Cu–H generated in situ did not proceed (entry 6). Next, conditions established by Lipshutz et al. were examined and expected to result a higher diastereoselectivity using the asymmetric ligand, but no selectivity was observed (entry 7) [51]. The reaction with NaBH₄ in the presence of CoCl₂, which generated Co–H, gave the desired product in 83% yield, but no selectivity was observed (entry 8) [50].

Table 18.2 1,4-Reduction of butenolides 29a and 29b

Unfortunately, we could not identify conditions for diastereoselective 1,4-reduction of the butenolide. Additionally, it was difficult to separate the diastereomeric mixture of the reduced products 13a and 13b. Therefore, we reconsidered the synthetic route to establish a more efficient route.

18.2.4 Second-Generation Approach to All-Cis Cyclopropane Using Iridium-Catalyzed Double Bond Isomerization

Although it was problematic to convert butenolides 29a and 29b synthesized from all-cis cyclopropane 31, the all-cis structure was selectively constructed by the palladium-catalyzed reduction of allyl carbonate esters 24d. These results suggested that the all-cis structure could be selectively constructed via a formation of the metal complex intermediate by approaching the bulky metal catalyst from the convex face of the bicyclic ring system. Thus, we focused on the iridium-catalyzed double bond isomerization in which the oxygen-functional group works as a directing group [52, 53]. In other words, the all-cis cyclopropane 35 could be constructed by the iridium-catalyzed double bond isomerization of alkylidenecyclopropane 24c through an intermediate in which a bulky iridium hydride approaches from the convex face of the molecule to form the complex. To avoid going through the butenolide structure (i.e., 29a and 29b) enroute to lactone 13, 1,3-diol 34 was set as an intermediate. This intermediate could be converted to nitrile 33 by distinguishing the reactivities of the two hydroxy groups (Scheme 18.13). Then, 1,3-diol 34 could be synthesized by introduction of a hydroxymethyl group at the α-position of aldehyde in 35.

Scheme 18.13

Retrosynthesis for the iridium-catalyzed double bond isomerization

The iridium-catalyzed double bond isomerization using a directing group was investigated using 22a, 23a, 23b, 24b, and 24c. The isomerization did not proceed when nitrile 22a was treated under a hydrogen atmosphere in the presence of Crabtree’s catalyst ([Ir(cod)(pyr)(PCy₃)]PF₆) (Scheme 18.14a). These results suggested that the cyano group contributed to inactivation rather than working as a directing group. On the other hand, when alcohol 23a was treated under the same conditions, the isomerization proceeded smoothly to give enol ether 37a as a single diastereomer. Similarly, the reaction of 23b having a TIPS group gave the all-cis cyclopropane 37b in 92% yield as a single diastereomer. However, further derivatization of 37a and 37b was difficult because side reactions, including ring-opening of the cyclopropane, occurred rather than deoxygenation.¹ Therefore, we examined the double bond isomerization of a deoxygenated substrate 24b. The reaction under a hydrogen atmosphere in the presence of Crabtree’s catalyst gave 38d in low yield (Scheme 18.14b). To improve the reactivity of the iridium catalyst, [Ir(cod)(pyr)(PCy₃)]BAr_F having a non-coordinating counter anion (tetrakis[3,5-bis(trifluoromethyl)phenyl]borate [BAr_F]) was used. Although the starting material was completely consumed, only the undesired trans-substituted cyclopropanes 38dʹ and 38dʺ were obtained. On the other hand, the reaction of allyl alcohol 24c with Crabtree’s catalyst gave the desired all-cis cyclopropane as a major product, albeit with low selectivity (2.7:1). The diastereoselectivity improved to 10:1 when using [Ir(cod)(pyr)(PCy₃)]BAr_F (Scheme 18.14c).

Scheme 18.14

Investigation of iridium-catalyzed double bond isomerization

The selectivity of the double bond isomerization can be rationalized as follows. When alcohols 23a and 23b are used as substrates, the hydroxymethyl group coordinated to the iridium center (i.e., intermediate E) and the C–H bond formed from only one side, which gave the all-cis cyclopropane (Scheme 18.15a). In the case of 24b, which had a methyl group substituent instead of a hydroxymethyl group, we speculated that double bond isomerization using [Ir(cod)(pyr)(PCy₃)]PF₆ (Crabtree’s catalyst) proceeded without coordination to iridium center. This was primarily because of the lower coordinating abilities of the oxygen-functional groups, such as TIPS and PMB ether, compared with the hydroxy group. This resulted in production of the isomerized product 38d in low yield with moderate diastereoselectivity. On the other hand, the trans-cyclopropane 38dʹ could be obtained when [Ir(cod)(pyr)(PCy₃)]BAr_F, which was susceptible to coordination, was used. The PMB ether, which was less sterically hindered than the TIPS ether, served as a coordinating group to give the trans isomer through intermediate F (Scheme 18.15b). The high ratio of the reduced product 38dʺ was attributed to the ease of hydrogenating the resulting disubstituted olefin of trans-substituted cyclopropane because of its lower steric hindrance. When allyl alcohol 24c was used as a substrate, the metal center of [Ir(cod)(pyr)(PCy₃)]BAr_F formed a strong coordination bond with the hydroxy group instead of the PMB ether, and then steric repulsion with the ligand resulted in formation of all-cis cyclopropane 35 through intermediate I (Scheme 18.15c). We also speculated that when Crabtree’s catalyst, which had weaker coordination ability than [Ir(cod)(pyr)(PCy₃)]BAr_F, was used, the isomerization proceeded without coordination and led to low cis-selectivity.

Scheme 18.15

Discussion of the cis/trans selectivity

Next, aldehyde 35 was converted to 1,3-diol 34 through aldol reaction with formaldehyde, 1,2-reduction of the resulting α,β-unsaturated aldehyde, and hydroboration–oxidation of an exo-methylene (Scheme 18.16). Diol 34 was then treated with DDQ to try to differentiate one hydroxy group from the other to obtain p-methoxybenzylidene acetal 39. Unfortunately, 39 was not obtained at all. Instead, tetrahydropyran 40a was obtained, albeit in low yield.

Scheme 18.16

Initial trial of the differentiation of diol 34

Despite the unexpected formation of the ether ring, we were able to distinguish one hydroxymethyl group of 1,3-diol 34 selectively albeit low yield. Thus, we investigated an optimization of this reaction (Table 18.3). Initially, considering the possibility that DDQ acted as an oxidant in the reaction, several copper catalysts were examined. The reaction did not proceed at all in the case of Cu(OAc)₂, while the reaction with Cu(OTf)₂ yielded a trace amount of the cyclized product 40b (entries 1 and 2). In sharp contrast, use of Cu(ClO₄)₂, which has a high Lewis acidity, dramatically improved the yields of 40a and 40b (entry 3). Next, we examined the Lewis acids Zn(OTf)₂, Sc(OTf)₃, and BF₃·OEt₂. The cyclized products 40a and 40b were obtained in approximately 80% combined yield when BF₃·OEt₂ was used (entries 4–6). We also obtained 40a and 40b quantitatively using pTsOH as a Brønsted acid (entries 7). Considering these results, we speculated that pTsOH or BF₃·OEt₂ effectively eliminated the PMBO group. Finally, the reaction with pTsOH was conducted in the presence of an excess amount of PhSH to capture the oxonium cation derived from the PMB group. This gave alcohol 40a in 88% yield in a chemoselective manner (entry 8).

Table 18.3 Differentiation of diol 34

In this reaction, p-methoxybenzyl alcohol 41 and bis(p-methoxybenzyl) ether 42 were obtained as by-products (Scheme 18.17). From these results, we hypothesized that the PMB ether 40b was formed through the following steps. First, elimination of the p-methoxybenzyl ether of 34 by acid-activation, which gave oxonium cation J and triol K, was followed by the formation of carbocation L-1 by elimination of a hydroxy group. Alternatively, the p-methoxybenzyloxy group of 34 could be directly eliminated to produce L-1. This secondary cation L-1 would be stabilized by the σ-donation of the cyclopropyl group [54]. The cation of L-1 was then reacted with the intramolecular hydroxy group to give tetrahydropyran 40a. Finally, the reaction of 40a with oxonium cation J led to the formation of PMB ether 40b.

Scheme 18.17

Plausible reaction mechanism

The reaction proceeded in a stereoselective manner at C8. The newly formed stereochemical configuration was determined by NOESY experiments of 40c, which was obtained via silylation of 40a. The stereoselectivity could be rationalized as follows. Although two conformations, L-1 and L-2, are possible for this cyclization, the methylene moiety of one of the hydroxymethyl groups of L-2 would experience steric repulsion with the hydrogen on C4 and the methyl group of C5 on the six-membered ring. Consequently, conformation L-1 was favored, leading to the formation of 40a. Although this reaction might be reversible, 40a would be thermodynamically more stable than 8-epi-40a because of similar steric repulsion.

The C–H oxidation was then examined for the ring-opening of the THP ring after formation of benzoyl ester from 40a. Oxidation using stoichiometric amounts of CrO₃ or a combination of RuCl₃ catalyst and NaIO₄ gave undesired lactone 45 and carboxylic acids 46 because of oxidation of a methylene instead of a methine (Table 18.4, entries 1 and 2) [55, 56]. In the case of (S,S)-Fe(pdp), reported by Chen and White, the reaction gave the desired keto alcohol 44 regioselectively in 65% yield, but required stoichiometric amounts of iron complexes (entry 3) [57]. When oxidation with dimethyldioxirane (DMDO) was attempted, the reaction did not complete even after 24 h although 44 was obtained in 22% yield. On the other hand, treatment with trifluoromethyl(methyl)dioxirane (TFDO) at 0 °C resulted in low regioselectivity because of the high reactivity of TFDO (entries 4, 5) [58]. Finally, when using TFDO at − 78 °C, regioselective C–H oxidation proceeded to give 44 in 96% yield (entry 6).

Table 18.4 C–H oxidation of tetrahydropyran

Next our attention turned to constructing the C-ring moiety to complete the total synthesis of avenaol (1).² The keto alcohol 44 was elongated through mesylation and S_N2 substitution with cyanide (Scheme 18.18). After DIBAL-H reduction of the resulting nitrile 33, hydrolysis under basic conditions was followed by treatment with an acid to yield a mixture of lactone 13b and alkene 47. This mixture was directly treated with pTsOH, resulting in the conversion of 13b to 47. Dihydroxylation of 47 gave a diol in a diastereoselective manner because OsO₄ was approached from the convex face. After regioselective silylation, the resulting triethylsilyl (TES) ether 48 was treated with methyl formate in the presence of potassium tert-butoxide to afford the formylated product [19]. Coupling of the product with bromobutenolide 11a afforded enol ether 49, which is a core structure of avenaol, as a diastereomeric mixture with the 2ʹ-epimer. Dess–Martin oxidation of 49 gave protected avenaol 50 and 2ʹ-epi-50, which were separated by silica gel column chromatography. The alcohol was oxidized after the formylation and introduction of the butenolide because cyclization via intramolecular aldol reaction during formylation and butenolide introduction would occur if it was oxidized first. Finally, treatment of 50 with HF·pyridine completed the total synthesis of avenaol (1). The spectral data, including ¹H, ¹³C NMR, FTIR, and ESI-MS spectra, of the synthetic avenaol were identical to those of the natural product reported by Yoneyama et al. [2]. The epimer of avenaol, 2ʹ-epi-1, was also synthesized by removing the TES group from 2ʹ-epi-50 by treatment with HF·pyridine. The relative configuration of 2ʹ-epi-1 was determined by X-ray crystallography. These results indicate that the structure of avenaol and its relative configuration are correct, albeit indirectly.

Scheme 18.18

End game for the total synthesis of avenaol

18.3 Synthesis of (+)-Avenaol from a Racemic Synthetic Intermediate

After achieving the total synthesis of racemic avenaol, we focused on its asymmetric synthesis. If the synthetic intermediate 18b, obtained by the asymmetric nucleophilic addition as shown in Scheme 18.3, was derivatized without epimerization through the established synthetic route, the optically active avenaol could be accessed. However, the asymmetric nucleophilic addition required a stoichiometric amount of a chiral ligand whose use is restricted by law. This method might be inefficient for the synthesis of enantiomers of the ABC ring moiety because the optically active ent-18b needed to be prepared in an early stage of the synthesis. Additionally, the introduction of the D-ring moiety did not occur stereoselectively. It has been reported that SL with an R configuration on C2ʹ of the D ring, which is the same as in the natural product, exhibits more germination-stimulating activity than that having a S configuration [59]. Therefore, we envisioned that the enantioselective introduction of the D-ring moiety using racemic synthetic intermediates would both enable efficient synthesis of optically active avenaol and provide a highly general method for the asymmetric synthesis of SLs. After various investigations, we found that a SL structural analog with a 1-indanone skeleton, enol 51a, reacted with racemic chlorobutenolide 11b in the presence of cesium carbonate and chiral thiourea-ammonium salt catalyst PTC-1 in chlorobenzene–water (20:1) to give 52a in 72% yield with an enantiomeric ratio (er) of 94:6 (Table 18.5, entry 1) [60]. This reaction could apply to enol 51b with two methyl groups at C5 and C6 and enol 51c fused at C6 and C7, which is a bulkier substrate (entries 2 and 3). Furthermore, diastereoselective introduction of the D-ring moiety into enantiomerically pure 51d was possible to form the artificial SL GR24 (52d, entry 4).

Table 18.5 Enantioselective acetalization for introduction of the D-ring butenolide

The optimized reaction conditions could be applied to the diastereomeric acetalization of several racemic enols that are the SL precursors. Racemic enol 51d was converted to (+)-(3aR,8bS,2ʹR)-52d (GR24, 54% yield, 82:18 er) and its diastereomer (−)-(3aS,8bR,2ʹR)-52d (40% yield, 88:12 er) (Scheme 18.19). Similarly, the bicyclic SLs (+)-(3aR,6aS,2ʹR)-52e (GR7, 41% yield, 81:19 er) and (−)-(3aS,6aR,2ʹR)-52e (34% yield, 85:15 er) were obtained from enol (±)-51e.

Scheme 18.19

Synthesis of SLs by PTC-catalyzed stereoselective acetalization

Finally, this diastereoselective acetalization was applied to the asymmetric synthesis of avenaol. A key synthetic intermediate (±)-53 was reacted under the optimum conditions and yielded a mixture of enol ether 49 (45%) and ent-2ʹ-epi-49 (30%) on introduction of the D-ring moiety (Scheme 18.20). Dess–Martin oxidation of the mixture was followed by separation by silica gel column chromatography to give (+)-50 and (−)-ent-2ʹ-epi-50. The protected avenaol (+)-50 was treated with HF·pyridine to obtain avenaol (+)-1 in an 81:19 er. The circular dichroism spectrum of (+)-1 was consistent with the reported Cotton effect, which confirmed this compound had the same absolute configuration as the natural product [2]. Because the 2ʹ-(R) epimer would be obtained in the case of PTC-1, it was strongly suggested that the absolute configuration of the natural product was 2ʹ-(R) as in the proposed structure. Similarly, the enantioselectivity of (−)-ent-2ʹ-epi-1 obtained by removing the TES group of (−)-ent-2ʹ-epi-50 was 96:4.

Scheme 18.20

Synthesis of (+)-avenaol by PTC-catalyzed stereoselective acetalization

18.4 Conclusions

We achieved the first total synthesis of avenaol (1) using the following key reactions: (1) rhodium-catalyzed intramolecular cyclopropanation of allenes to synthesize alkylidenecyclopropanes, (2) iridium-catalyzed diastereoselective double bond isomerization to construct all-cis cyclopropane, (3) distinction of two hydroxymethyl groups on the 1,3-diol via intramolecular S_N1-type reactions, and (4) cleavage of the THP ring by regioselective C–H oxidation (Scheme 18.21). This is the first synthetic proof of the proposed relative stereo configuration of avenaol. After this study, we succeeded in the total synthesis of shagene A and B by extending the synthetic strategy of all-cis cyclopropanes developed in this study [61]. It is expected that these synthetic strategies, including other attempts such as hydrogenation, radical cyclization, and palladium-catalyzed reduction of alkylidenecyclopropane derivatives, described in this manuscript will be widely used in the future. Furthermore, we have developed a stereoselective D ring introduction method for SL by acetal formation with γ-chlorobutenolide via chiral thiourea-quaternary ammonium salt-catalyzed dynamic kinetic resolution. By applying the developed reaction conditions to racemic substrates, optically active SLs could be readily synthesized. Finally, we succeeded in synthesizing optically active avenaol and confirming its absolute configuration. This method is expected to be used for various substrates as a stereoselective introduction of the strigolactone D-ring moiety.

Scheme 18.21

Overview of strategies for access to all-cis cyclopropanes and total synthesis of avenaol