Total Syntheses of 11-Acetoxy-4-deoxyasbestinin D, 4-Deoxyasbestinin C, Asbestinin-10, -20, -21 and -23.

Six members of the asbestinin family of marine diterpene natural products have been synthesized in an efficient and stereoselective manner from a single oxa-bridged intermediate. Five of these natural products have not been synthesized previously and the structures of four of them have been confirmed as those proposed originally or following revisions to the original structures. The fifth natural product-asbestinin-21-has been shown to be a diastereomer of the compound that had been proposed previously.

Introduction

The asbestinins are structurally complex tetracyclic marine diterpenes that have been isolated from the gorgonian octocorals Briareum asbestinum and Briareum polyanthes. The isolation and characterization of the first members of the family—asbestinins‐1–5—were reported by Faulkner, Clardy and co‐workers in 1980.1 The 39 asbestinins that have been isolated and characterized to date partition into two distinct structural groups: members that lack a C‐4 substituent, such as 11‐acetoxy‐deoxyasbestinin D (1) and 4‐deoxyasbestinin C (2), and those that are oxidised at C‐4, such as asbestinin‐12 (3) and asbestinin‐2 (4) (Figure 1).2, 3, 4, 5, 6, 7, 8 A biosynthetic route to the asbestinins from related briarellin natural products has been proposed in which Wagner–Meerwein rearrangement transfers a methyl group from C‐11 to C‐12.1, 9 The briarellin natural products are, in turn, thought to be derived from the simpler cladiellins (eunicellins).

Figure 1

Examples of asbestinin natural products with and without an oxygen‐containing substituent at the C‐4 position.

Many of the asbestinins have been reported to possess significant biological activities: 4‐deoxyasbestinin A, 11‐acetoxy‐4‐deoxyasbestinin B, 4‐deoxyasbestinin C, 11‐acetoxy‐4‐deoxyasbestinin D and asbestinins‐6–10 possess significant in vitro activity against several cancer cells lines; nor‐asbestinin A, 11‐acetoxy‐4‐deoxyasbestinin F, and asbestinins‐10, ‐20, ‐21 and ‐26, display activity against the malarial parasite Plasmodium falciparum; 4‐deoxyasbestinin A, 11‐acetoxy‐4‐deoxyasbestinin B, 4‐deoxyasbestinin C, 11‐acetoxy‐4‐deoxy‐asbestinin D are active against the bacterium Klebsiella pneumoniae. However, the biological activities of most of the asbestinins have not been fully determined.10, 11

Several of the structures that were proposed for the asbestinins upon isolation have been revised as a consequence of the acquisition of better quality NMR data and the advent of new NMR techniques.8 Several asbestinins that were thought to be oxidized at the C‐4 position are likely to be 4‐deoxyasbestinins that contain a hydroxyl, hydroperoxy or carbonyl group at C‐6 instead (Figure 2). This work has resulted in tentative revision of the structures of asbestinin‐20 (5→6), 11‐acetoxy‐4‐deoxyasbestinin F (9→10), asbestinin‐10 (13→14) and asbestinin‐21 (15→16). It is likely that the structures of 4‐deoxyasbestinin G (7→8), asbestinin‐9 (11→12), asbestinin‐22 (17→18) and asbestinin‐23 (19→20) also require revision.8

Figure 2

Asbestinin natural products that have had their structures corrected or are likely to require structural revision.

The asbestinins possess a complex fused tetracyclic core that contains two medium‐sized cyclic ethers (ring sizes seven and nine) and bears a total of 9–12 stereogenic centres. The significant synthetic challenges presented by the asbestinins combined with their biological activities and the fact that there are ambiguities about their structures in many cases, makes them highly alluring targets for total synthesis. To date, 11‐acetoxy‐4‐deoxyasbestinin D and asbestinin‐12 are the only members of the asbestinin family of natural products that have been synthesized, as reported by Crimmins and Ellis in 2005 and 2008.12

Results and Discussion

At the outset, we intended to synthesize 11‐acetoxy‐4‐deoxyasbestinin D (1) and 4‐deoxyasbestinin C (2), and prepare several other asbestinins from these natural products thereafter.3 The retrosynthetic analysis of the natural products 1 and 2 began with the cleavage of the ester and formation of a ketone at C‐11 and removal of the C‐12 methyl substituent to give advanced tetracyclic intermediate i (Scheme 1). Sequential opening of the oxepane by scission of the C−O bond distal to the quaternary centre at C‐3, conversion of the methyl substituent into a methylene group and the C‐11 ketone into an enol ether, followed by removal of the C‐3 methyl substituent led to the tricyclic ketone ii. Subsequent conversion of the side‐chain allylic ether into a methyl ketone produced intermediate iii, an analogue of a synthetic intermediate used in our total syntheses of members the cladiellin natural products.13 Further Diels–Alder disconnection led to the diene iv, and disconnection of the diene revealed the bridged bicyclic ether v.

Scheme 1

Retrosynthetic analysis of 11‐acetoxy‐4‐deoxyasbestinin D (1) and 4‐deoxyasbestinin C (2).

The ketone corresponding to tricyclic intermediate iii in Scheme 1 was prepared as shown in Scheme 2. Several approaches to the synthesis of the bridged bicyclic ketone 23 (a compound corresponding to ketone v in Scheme 1) from tetrahydropyranyl ester 21 were explored. In the first approach, we employed the method that we had used in our cladiellin syntheses.13 Conversion of ester 21 into α‐diazo ketone 22 proceeded smoothly and subsequent sequential copper‐catalysed carbenoid generation, ylide formation and rearrangement delivered the oxa‐bridged bicyclic ketone 23 in high yield. In the second approach, the metal‐mediated reaction of triazole 25 according to conditions described by Boyer was explored.14 Ester 21 was subjected to reduction to give the corresponding aldehyde followed by Bestmann–Ohira homologation to deliver alkyne 24.15 Deprotonation of alkyne 24 and reaction of the resulting anion with tosyl azide afforded triazole 25.14 The rhodium‐mediated reaction of triazole 25 produced the expected ketone 23 with a high level of diastereocontrol but in lower yield than that obtained from the copper‐catalysed reaction of α‐diazo ketone 22. Finally, direct oxidative gold‐catalysed cyclization of alkyne 24 was performed in an attempt to generate the bicyclic oxonium ylide directly from this substrate and thus obviate the use of either α‐diazo ketone 22 or triazole 25.16 Unfortunately, ester 26 was obtained instead of the required rearrangement product 23.

Scheme 2

Oxonium ylide formation and ring expansion to form the oxa‐bridged tricyclic core system. a) LiOH, EtOH‐H2O (3:1), RT, 86 %; b) iBuOCOCl, Et3N, RT, then CH2N2, Et2O, RT, 82 %; c) Cu(hfacac)2, CH2Cl2, reflux, 90 % (5:1, Z:E); d) iBu2AlH, CH2Cl2, −78 °C; e) MeCOC(N2)P(O)(OMe)2, K2CO3, MeOH, RT, 86 % (2 steps); f) nBuLi. TsN3, THF, −78 °C, 87 %; g) Rh2(OAc)4, DCE, reflux, 48 % (>20:1, Z:E); h) SPhosAuNTf2, ClCH2CO2H, 5‐bromo‐2‐methyl‐pyridine‐N‐oxide, DCE, RT, 36 % (26).

Bicyclic ether 23 was converted into tricyclic ketone 27, as reported previously.13 Further functionality was introduced to enable the construction of the oxepane, the final ring in the tetracyclic ensemble (Scheme 3). One‐carbon chain extension was performed by conversion of the methyl ketone into triflate 28 followed by palladium‐mediated carbonylation in methanol to deliver ester 29 in excellent yield. Selective reduction of the carbonyl group of the α,β‐unsaturated ester was accompanied by the loss of the TBS (tert‐butyldimethylsilyl ether) protecting group, and diol 30 was produced in high yield. Selective acetylation of the primary allylic alcohol and subsequent oxidation of the free secondary hydroxyl group delivered ketone 31. Treatment of this ketone with a large excess of methylmagnesium chloride resulted in a highly stereoselective installation of the C‐3 methyl substituent and concurrent cleavage of the acetate group.17 The enol ether was converted into the corresponding C‐11 ketone by hydrolysis during the acidic workup and keto diol 32 was produced in excellent yield.

Scheme 3

Functionalised of the tricyclic core in preparation for construction of the oxepane. a) NaN(SiMe3)2, PhNTf2, THF, −78 °C; b) Pd(PPh3)4, CO, MeOH, iPr2NEt, RT, 96 % (2 steps); c) iBu2AlH, PhMe, −35 °C, 80 %; d) Ac2O, DMAP (4‐dimethylaminopyridine), Et3N, CH2Cl2, RT; e) Dess–Martin periodinane, pyridine, CH2Cl2, RT, 98 % (2 steps); f) MeMgCl, THF, 0 °C→RT then 1 m HCl aq., RT, 87 %.

Conversion of tricyclic diol 32 into the fully functionalised tetracyclic asbestinin framework was accomplished by following the route shown in Scheme 4. Oxepane formation was performed by treatment of diol 32 with triflic anhydride in the presence of 2,6‐lutidine. The primary allylic triflate was not isolated from this reaction and instead the required cyclic allylic ether 33 was obtained in excellent yield as a result of immediate intramolecular nucleophilic displacement of the triflate by the tertiary alcohol. Hydrogenation of allylic ether 33 in the presence of Adams’ catalyst was both chemoselective and stereoselective; the tetracyclic ketone 34 was obtained as a single diastereomer in high yield.18 To convert ketone 34 into the natural products 1 and 2 it was necessary to install the C‐12 methyl substituent, reduce the ketone and esterify the resulting alcohol. Direct methylation at C‐12 by ketone deprotonation and alkylation was low‐yielding and non‐stereoselective, so other methods for the introduction of the methyl substituent were explored. Reaction of ketone 34 with Bredereck's reagent19 at 100 °C in DMF produced enamine 35, which was subjected to immediate reduction with DIBAL‐H to afford enone 36. Reduction of this enone under Luche conditions produced the crystalline allylic alcohol 37 as a single diastereomer, but the configuration at the hydroxyl‐bearing stereogenic centre (C‐11) was not the one required for the synthesis of the asbestinins, as revealed by X‐ray analysis.20, 21 Treatment of enone 36 with either L‐Selectride or Stryker's reagent22 resulted in the formation of a diastereomeric mixture of the required ketone 38 a and its C‐12 epimer 38 b, in which the latter predominated marginally. The epimeric ketones were separated and equilibration of each ketone under mildly basic conditions delivered a mixture of diastereomers in a similar ratio to that obtained by conjugate reduction of enone 36.

Scheme 4

Completion of the tetracyclic asbestinin core. a) Tf2O, 2,6‐lutidine, CHCl3, 0 °C, 97 %; b) H2, PtO2, EtOAc, RT, 90 %; c) tBuOCH(NMe2)2, DMF, 100 °C; d) iBu2AlH, CH2Cl2, −78 °C then MeI, RT, 52 % (2 steps); e) NaBH4, CeCl3⋅7 H2O, MeOH, −78 °C, 49 % (3 steps); f) L‐Selectride, THF, −78 °C, 76 % (30 % 38 a+46 % 38 b); g) [CuH(PPh3)]6, PhMe, −78 °C, 62 % (25 % 38 a+37 % 38 b) (3 steps); h) K2CO3, MeOH, RT, 85–99 % (1:1.4, 38 a:38 b).

Completion of the total syntheses of 11‐acetoxy‐4‐deoxyasbestinin D (1) and 4‐deoxyasbestinin C (2) was finally possible (Scheme 5). Reduction of ketone 38 a to give the alcohol by treatment with sodium borohydride was highly diastereoselective.23 Immediate acetylation of the resulting alcohol 39 provided 11‐acetoxy‐4‐deoxyasbestinin D (1) in excellent yield and reaction of alcohol 39 with butyric anhydride produced 4‐deoxyasbestinin C (2) in good yield. The NMR and other characterization data of the synthetic esters 1 and 2 matched the data reported for each natural product.

Scheme 5

Completion of the syntheses of 11‐acetoxy‐4‐deoxyasbestinin D (1) and 4‐deoxyasbestinin C (2). a) NaBH4, CH2Cl2‐MeOH (1:1), RT; b) Ac2O, DMAP, Et3N, CH2Cl2, RT, 88 % (2 steps); c) (n‐PrCO)2O, DMAP, Et3N, CH2Cl2, RT, 50 % (2 steps).

Several other asbestinin natural products were synthesized from 11‐acetoxy‐4‐deoxyasbestinin D (1) in just two or three steps. Of particular interest was the synthesis of asbestinin‐20 (6) to verify its reassigned structure (Scheme 6). 11‐Acetoxy‐4‐deoxyasbestinin D (1) was first epoxidized by treatment with mCPBA to give a mixture (2:1) of diastereomeric epoxides 40 a and 40 b. The crystalline epoxides were readily separable by chromatography and their structures were determined by X‐ray crystallography.20 To amplify or override the weak substrate bias and thereby generate epoxides 40 a and 40 b with higher levels of diastereocontrol, the epoxidation reaction was performed by the application of Shi's asymmetric protocol using both enantiomers of the fructose‐derived catalyst.24 In the case of the (+)‐catalyst, the relatively weak substrate bias was reinforced and the ratio of diastereomers improved to 10.5:1 (40 a:40 b). In the mismatched case, in which the (−)‐catalyst was employed, the substrate bias was overturned and a 1:5.5 ratio of diastereomers (40 a:40 b) was obtained. Epoxides 40 a and 40 b were then converted into the corresponding allylic alcohols by sequential reaction with tert‐butyldimethylsilyl triflate in the presence of 2,6‐lutidine and treatment with TBAF.25 The crystalline allylic alcohols 6 and 9 were obtained with yields of 39 and 64 %, respectively; their structures were confirmed by X‐ray crystallography.20 In the case of epoxide 40 a, treatment with titania‐supported gold nanoparticles in dichloroethane (DCE) at 80 °C, according to the procedure described by Garcia and Stratakis,26 delivered allylic alcohol 6 in 35 % yield. Comparison of NMR and other characterization data of alcohols 6 and 9 with that of natural asbestinin‐20 confirmed that the natural product is compound 6. Thus, the reassigned structure and relative stereochemistry of asbestinin‐20 published by Ospina and Rodríguez is correct.8 Dess–Martin oxidation of asbestinin‐20 (6) produced asbestinin‐10 (14) and the structure of this natural product was confirmed. Enone reduction under Luche conditions afforded the allylic alcohols in a 3.2:1 ratio favouring asbestinin‐20 (6), a result consistent with that reported by Ospina and Rodríguez when they reduced natural asbestinin‐10 (14) under similar conditions.8

Scheme 6

Synthesis of asbestinin‐20 (6), 6‐epi‐asbestinin‐20 (9) and asbestinin‐10 (14). a) mCPBA, CH2Cl2, −10 °C, 82 % (59 % 40 a+23 % 40 b); b) (+)‐Shi epoxidation catalyst, oxone, K2CO3, nBu4NHSO4, Na2B4O7×10H2O, MeCN‐DMM (1:2) 0 °C, 81 % (75 % 40 a+6 % 40 b); c) (−)‐Shi epoxidation catalyst, oxone, K2CO3, nBu4NHSO4, Na2B4O7×10H2O, MeCN‐DMM (1:2) 0 °C, 84 % (13 % 40 a+71 % 40 b); d) tBuMe2SiOTf, 2,6‐lutidine, CH2Cl2, 0 °C then TBAF, THF, RT, 39 % (6), 64 % (9); e) Au‐TiO2, DCE, 80 °C, 35 %; f) Dess–Martin periodinane, pyridine, CH2Cl2, RT, 75 %; g) NaBH4, CeCl3⋅7 H2O, CH2Cl2‐MeOH (1:1), −78 °C, 84 % (64 % 6+20 % 9).

The next target to be synthesized was asbestinin‐21, which had been originally identified as the ketone 15 (Figure 2), but was later reassigned to be the α‐hydroxy ketone 16 (Scheme 7).6, 8 Dihydroxylation of 11‐acetoxy‐4‐deoxyasbestinin D (1) under standard Upjohn dihydroxylation conditions afforded a diastereomeric mixture of 1,2‐diols 41 a and 41 b in 56 % yield, in which the crystalline diol 41 a predominated (5:1 ratio).20 The diol 41 a was separated from the minor isomer 41 b and oxidized to give α‐hydroxy ketone 16 by the use of the Dess–Martin protocol. Alternatively, the diastereomeric mixture of diols 41 a and 41 b was oxidized directly to produce a separable mixture of α‐hydroxy ketones 16 (67 % yield) and 42 (14 % yield).

Scheme 7

Stereoselective dihydroxylation of 11‐acetoxy‐4‐deoxyasbestinin D (1) and synthesis of asbestinin‐21 (42) and 7‐epi‐asbestinin‐21 (16). a) OsO4, NMO (N‐methylmorpholine N‐oxide), THF‐H2O (1:1), RT, 56 % (5:1, 41 a:41 b); b) Dess–Martin periodinane, pyridine, CH2Cl2, RT, 87 % (41 a→16), 81 % (41 a+41 b→16+42); c) AD‐mix‐α, tBuOH‐H2O (1:1), 0 °C, 23 % (>12:1, 41 a:41 b)+46 % (16); d) AD‐mix‐β, tBuOH‐H2O (1:1), 0 °C, 75 % (8:1, 41 a:41 b)+10 % (16).

Attempts were made to produce diol 41 b as the major diastereomer by performing dihydroxylation according to the Sharpless asymmetric protocol.27 The use of AD‐mix‐α to perform the dihydroxylation reaction of 11‐acetoxy‐4‐deoxyasbestinin D (1) resulted in the formation of diol 41 a with an enhanced level of diastereocontrol (>12:1). However, diol 41 a was obtained in only 23 % yield, and the major product was the α‐hydroxy ketone 16 (46 % yield), an unexpected product resulting from over‐oxidation. In the case where substrate control and reagent control were expected to be mismatched, dihydroxylation of alkene 1 with AD‐mix‐β produced a mixture of diols 41 a and 41 b (75 % yield combined) along with a small amount (10 % yield) of α‐hydroxy ketone 16. Interestingly, the dihydroxylation reaction proceeded with a higher level of diastereocontrol (8:1) than for the case when the reaction was performed under simple Upjohn conditions. This unexpected result presumably reflects the larger steric bulk of the osmium complex in the case of AD‐mix and the fact the reaction proceeds under substrate control rather than reagent control. Comparison of the 1H and 13C NMR data of the α‐hydroxy ketones 16 and 42 with those of natural asbestinin‐21 revealed that the natural product is 42 rather than 16, a finding which established that the configuration of the hydroxyl‐bearing the C‐7 stereogenic centre had been misassigned by Ospina and Rodríguez when performing their structural revision.8

The last natural product to be synthesized in our study was asbestinin‐23.6, 8 Reaction of 11‐acetoxy‐4‐deoxyasbestinin D (1) with borane‐THF complex followed by oxidative work‐up produced an inseparable diastereomeric mixture of alcohols (Scheme 8). This mixture of alcohols was subjected to immediate oxidation with Dess–Martin periodinane to give ketones 20 and 43, which could be separated and were isolated in yields of 20 and 49 %, respectively. Reduction of ketone 43 with L‐Selectride afforded alcohol 44, a crystalline solid, in a highly stereoselective manner (>20:1 selectivity). The configurations of the stereogenic centres at C‐6 and C‐7 of alcohol 44 were determined by X‐ray crystallography,20 and the configuration of the C‐7 stereocentres of ketones 43 and 20 were established consequently. Comparison of NMR data of ketones 20 and 43 with the data of the natural asbestinin‐23 revealed that the natural product is ketone 20, which was suggested by Ospina and Rodríguez as the most likely alternative to the originally assigned structure (19, Figure 2).8

Scheme 8

Completion of the syntheses of asbestinin‐23 (20) and 7‐epi‐asbestinin‐23 (43). a) BH3⋅THF, THF, 0 °C then NaBO3⋅4 H2O; b) Dess–Martin periodinane, pyridine, CH2Cl2, RT, 69 % (2 steps) (49 % 43 a, 20 % 20); c) L‐Selectride, THF, −78 °C, 64 % (>20:1).

Conclusions

We have completed the stereoselective total syntheses of six members of the asbestinin family of marine diterpene natural products from the bridged‐bicyclic ether 23, which had been prepared previously from readily available starting materials. Syntheses of 11‐acetoxy‐deoxyasbestinin D (1) and 4‐deoxyasbestinin C (2) have been completed in 16 steps from the bicyclic ketone 23 and the former has been converted into abestinin‐10, abestinin‐20, abestinin‐21 and abestinin‐23 in two or three additional steps. The syntheses of abestinin‐10, abestinin‐20, and abestinin‐23 confirm the revised structures for these compounds proposed by Ospina and Rodríguez in 2006. In the case of abestinin‐21, our work has shown that this compound is the C‐7 epimer of the revised structure proposed by Ospina and Rodríguez.

Conflict of interest

The authors declare no conflict of interest.

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

Supplementary

Click here for additional data file.