Download PDF
Synthesis 2010(16): 2841-2852  
DOI: 10.1055/s-0030-1258151
FEATUREARTICLE
© Georg Thieme Verlag Stuttgart ˙ New York

Enantioselective Synthesis of (+)-Polyanthellin A via Cyclopropane-Aldehyde (3+2)-Annulation

Matthew J. Campbell, Jeffrey S. Johnson*
Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3290, USA
Fax: +1(919)9622388; e-Mail: jsj@unc.edu;

Further Information

Publication History

Received 4 June 2010
Publication Date:
07 July 2010 (online)

 PDF Download Permissions and Reprints All articles of this category

Biographical Sketches

Matthew Campbell was born in 1982 in Ft. Belvoir, Virginia, but was raised in western Pennsylvania. He received his B.S. degree in Chemistry from the University of Pittsburgh where he conducted research under Prof. Dennis P. Curran. He attended graduate school at the University of North Carolina, Chapel Hill, and received his Ph.D. in 2009 under the tutelage of Prof. Jeffrey S. Johnson in the areas of electrophilic amination and [3+2]-annulation chemistry. He is currently a postdoctoral associate at the University of California-Berkeley, working in the laboratories of Prof. F. Dean Toste toward developing new chiral counterions for use in gold catalysis.
Jeffrey Johnson received a B.S. degree in chemistry from the University of Kansas­ in 1994, graduating with Highest Distinction and Honors in Chemistry. He was an NSF Graduate Fellow in the laboratory of David Evans at Harvard University, where he earned his Ph.D. in 1999. He undertook an NIH Postdoctoral Fellowship in the laboratory of Robert Bergman at the University of California at Berkeley from 1999-2001 and began as an Assistant Professor at the University of North Carolina in July, 2001. He is now a Professor of Chemistry.

Abstract

The asymmetric synthesis of the cladiellin diterpene natural product (+)-polyanthellin A is described. The core tetrahydrofuran was constructed using a stereospecific and stereoselective (3+2)-annulation of a donor-acceptor cyclopropane and a labile β-silyloxy aldehyde. These particular reactants necessitated the application of a new Lewis acid, MADNTf2 [(ArO)2AlNTf2], to avoid competitive elimination. Ring-closing metathesis was employed to form the oxonane ring at C3-C4 and give a functional group handle that could be elaborated to the natural product.

Key words

annulation - total synthesis - heterocycles - Lewis acids - ring opening

Introduction

The closely related cladiellin, briarellin, and asbestinane natural product families are secondary metabolites produced exclusively by gorgonian octocorals (order Gorgonacea, phylum Cnideria), presumably derived from cembranes via various oxidative cyclizations. [¹] These compounds exhibit toxicity toward a broad array of aquatic organisms including mollusks, crustaceans, echinoderms, and fish. [²-4] Biological screening has revealed properties that may be of use in medical contexts, as certain members exhibit antitumor, anti-inflammatory, and antimitotic activities. Furthermore, the structural complexity of cyclized cembranoids has captured the interest of the synthetic community as they provide a testing ground for emerging methodologies. Beginning with Overman’s synthesis of (-)-7-deacetoxyalcyonin acetate, which utilized a Prins-pinacol reaction as the key step, [5] [6] myriad strategies have been developed to construct the hydroisobenzofuran core and the oxacyclononane ring, two distinguishing structural features contained by all these compounds. These previous total and partial syntheses have been reviewed. [7]

The synthetic strategies devised to prepare the cladiellin skeleton can be sorted into two approaches: those that initially target the hydroisobenzofuran, and those that first form the oxonane. The structural variety found throughout the 2,11-cyclized cembranoids has ensured that neither strategy has become dominant; however, the syntheses by Overman [5] [6] [8] and Hoppe [9] (vigulariol, 2) are noteworthy for their convergence, and both first construct the hydroisobenzofuran and subsequently install the oxonane. These syntheses distill into three modular components (two functionally rich aldehydes and a cyclohexene core), and use a Lewis acid catalyzed cyclization onto an in situ generated oxocarbenium ion. We proposed to build the core by using a (3+2) annulation between a donor-acceptor (D-A) cyclopropane and an aldehyde. Although the oxocarbenium ion is generated by nucleophilic substitution instead of condensation, the intermediate is reminiscent of that found in Hoppe’s synthesis (Scheme  [¹] ).

Scheme 1 Approaches to the cladiellin core using oxocarbenium intermediates

Polyanthellin A (1) contains a second ether bridge embedded within the oxacyclononane moiety, a rare structural attribute among cladiellin diterpenes shared only with vigulariol (Figure  [¹] ). The structure of sclerophytin A (3) was initially proposed to contain this feature, [¹0] [¹¹] though the structure was revised to that of 4 following independent syntheses by the Overman and Paquette groups. Regardless, these works established intramolecular alkoxymercuration as a viable strategy for constructing the tetrahydropyran. This approach was subsequently utilized by Kim and co-workers to complete the first total synthesis of (+)-polyanthellin A, [¹²] as well as by Molander to prepare the corresponding 3,7-diastereomer. Poly­anthellin A has been isolated as both the dextro- and levo­rotatory forms from the west Pacific ocean near Australia and the Caribbean off the coast of Puerto Rico, respectively. [¹³] [¹4] To our knowledge, it is the only cladiellin diterpene to carry this distinction. (-)-Polyanthellin A is reported to be an antimalarial agent. [¹4]

Figure 1 Cladiellin diterpenes containing two ether bridges

Our laboratory has previously described the Lewis acid catalyzed annulation between donor-acceptor cyclopropanes and aldehydes to produce 2,5-disubstituted tetra­hydro­furans with high levels of cis/trans diastereo­-selectivity. [¹5-¹7] The reaction proceeds with nearly complete transfer of chirality when enantioenriched cyclopropanes are used, resulting from inversion at the cyclopropane stereocenter during annulation (Scheme  [²] ). This result and subsequent studies support a mechanism in which nucleophilic attack of the aldehyde on the cyclopropane leads to an open-chain zwitterion. A diastereo­selective aldol reaction occurs rapidly to furnish the substituted tetrahydrofuran. To access the correct relative stereochemistry found within the polyanthellin hydroisobenzofuran core, we surmised that the donor substituent of the cyclopropane must be located on the more sterically encumbered α-face of the bicyclo[4.1.0]heptanone. Before devising our retrosynthesis of polyanthel­lin A, we prepared two model bicyclic cyclopropanes that would allow us to explore the feasibility of the annulation.

Scheme 2 Tin(II) triflate catalyzed (3+2) annulation of 5 with benz­aldehyde

Results and Discussion

The ethylcyclopropane 10 was synthesized in four steps, beginning with N-(trimethylsilyl)diethylamine-catalyzed conjugate addition of isovaleraldehyde to methyl vinyl ketone to furnish directly keto aldehyde 7 (Scheme  [³] ). [¹8] A chemoselective Wittig reaction provided (Z)-alkene 8, which was converted into keto ester 9 with Mander’s reagent. [¹9] Cyclopropane 10 was formed using a protocol developed by Yang [²0] that preserves the olefin geometry through the reaction, placing the ethyl group on the concave face of the bicycle.

The vinylcyclopropane 14 was prepared using a silyl ether as a latent olefin (Scheme  [³] ). This conservative approach was taken to avoid potential complications with a conjugated diene during cyclopropanation. Additionally, a methyl­ ester was chosen in preference to the ethyl ester to simplify the spectrum for eventual NOESY experiments. After successful Wittig olefination and carboalkoxylation, the Yang cyclopropanation [²0] of 12 was found to be ineffective, presumably an effect of the TBSO group. Instead, the cyclopropane was installed by conversion into the diazo compound and rhodium(II)-catalyzed cyclopropanation. Silyl ether removal with fluorosilicic acid afforded alcohol 13 in 50% yield over three steps. Grieco selenoxide elimination [²¹] provided the vinylcyclopropane 14. NOESY experiments confirmed the relative stereo­chemistry.

Scheme 3 Reaction conditions: (a) MVK, Et2NTMS (10 mol%), MeCN, reflux; (b) 1. PrPPh3Br, BuLi, THF, 0 ˚C, 2. 7; (c) 1. LDA, THF, -78 ˚C, 2. HMPA; EtOC(O)CN; (d) Mg(ClO4)2, I2, Et3N, CH2Cl2, r.t.; (e) 1. TBSO(CH2)3PPh3Br, BuLi, THF, 0 ˚C, 2. 7; (f) 1. LDA, THF, -78 ˚C, 2. HMPA; MeOC(O)CN; (g) p-ABSA, Et3N, MeCN; (h) Rh2(OAc)4 (3 mol%), CH2Cl2; (i) H2SiF6, MeCN; (j) 1. 2-O2NC6H4SeCN, Bu3P, THF, 2. H2O2.

Using the standard conditions developed in our laboratory for the (3+2) annulation, several cyclopropane/aldehyde combinations were found effective: alkyl-aryl, alkyl-alkyl, and vinyl-alkyl. The rate, diastereoselectivity, and efficiency of the annulations were all consistent with expectations founded from our experience with simpler systems: vinylcyclopropanes are more active than their alkyl counterparts, and aryl aldehydes give higher diastereomeric ratios than alkyl aldehydes (Scheme  [4] ). The lowest yielding annulation is the reaction of ethylcyclopropane 10 with isobutyraldehyde, requiring 30 mol% tin(IV) chloride at 45 ˚C for 14 hours to give only a 71% yield of 16 with 7:1 dr. Benzaldehyde reacted at room temperature with 20 mol% hafnium(IV) triflate to give 74% of the cycloadduct 15 in 20:1 dr. Cycloaddition of the activated vinylcyclopropane 14 with isobutyraldehyde was most facile and provided 17 in 93% yield with a moderate 10:1 dr. Single crystals suitable for X-ray diffraction analysis were obtained from the 2,4-dinitrophenylhydrazone of 16, and the structure was solved to confirm our preliminary stereochemical assignments based on NOESY experiments. [²²] Finding the relative stereochemistry of the adducts consistent with that needed to prepare polyanthellin A, we proceeded with planning our synthesis using the (3+2) strategy. [²³]

Scheme 4 Reaction conditions: (a) PhCHO, Hf(OTf)4 (20 mol%), CH2Cl2, r.t., 4 h; (b) i-PrCHO, SnCl4 (30 mol%), DCE, 45 ˚C, 14 h; (c) i-PrCHO, SnCl4 (10 mol%), CH2Cl2, r.t., 1.5 h.

To begin our retrosynthetic analysis of polyanthellin A (1), we reasoned that a double methylene Wittig and subsequent oxymercurations of protected hydroxy dione 18 would afford 1, in close analogy with Kim’s synthesis (Scheme  [5] ). This dione would arise from allylic oxidation, Krapcho decarboalkoxylation, and hydrogenation of oxonene 19, itself prepared by a ring-closing metathesis (RCM) of adduct 20. Our studies with model systems suggested a challenging allylcyclopropane/alkyl aldehyde annulation would be feasible, with the requisite cyclopropane 21 available using the methods employed to prepare the model bicycles. A suitably protected hydroxy aldehyde 22 would be formed by a vinylcuprate ring opening of an enantiopure epoxide and subsequent functional group manipulations.

Scheme 5 Retrosynthetic analysis of polyanthellin A (1)

The synthesis of allylcyclopropane 21 began with an enantioselective organocatalyzed conjugate addition of isovaleraldehyde to methyl vinyl ketone, as reported by Gellman (Scheme  [6] ). [²4] We found the use of catechol 26 as an additive was essential, and obtained the keto aldehyde 7 in 90% yield and >97:3 e.r. after 48 hours. [²5] Wittig olefination in tetrahydrofuran provided the skipped diene 23 in 59% yield. Changing the counterion to Na+ or K+ reduced the yield, but the addition of a co-solvent was beneficial. A THF-HMPA (2:1) mixture was found to be optimal (80%), but we chose to use DMPU as a safer, albeit slightly less effective alternative (69%). Keto ester 24 was prepared with Mander’s reagent. [¹9] Yang’s cyclopropanation conditions [Mg(ClO4)2, I2, Et3N]²0 failed in the presence of the terminal olefin, so the use of metal carbenoids were investigated. After diazo transfer with 4-acet­amidobenzenesulfonyl azide (p-ABSA), experiments revealed rhodium(II) octanoate dimer [Rh2(oct)4] in dichloromethane to be optimal for the cyclopropanation, providing 21 in 60% yield.

Scheme 6 Synthesis of allylcyclopropane 59. Reaction conditions: (a) MVK, 25 (5 mol%), 26 (20 mol%), neat, 4 ˚C; (b) 1. H2C=CH(CH2)2PPh3Br, BuLi, THF-DMPU (2:1), 2. 7, -78 to 0 ˚C; (c) 1. LDA, THF, -78 ˚C, 2. HMPA, 3. MeOC(O)CN; (d) p-ABSA, Et3N, MeCN; (e) Rh2(oct)4 (0.5 mol%), CH2Cl2.

Hexenal 30 was prepared uneventfully from methallyl alcohol in six steps (Scheme  [7] ). Sharpless asymmetric epoxidation [²6] [²7] provided the known epoxy alcohol 27 (er >97.5:2.5), [²8] [²9] which was opened with a vinylcuprate to give diol 28. [³0] The primary alcohol was selectively converted into the nitrile by activation as a sulfonate and SN2 displacement by cyanide. The tertiary alcohol was protected to afford the TMS ether 29. The requisite aldehyde 30 was prepared in 86% yield by diisobutylaluminum hydride reduction and silica gel promoted hydrolysis of the intermediate imine.

Scheme 7 Synthesis of hexenal 30. Reaction conditions: (a) Ti(Oi-Pr)4 (7.5 mol%), (-)-DET (10 mol%), TBHP, 4 Å MS, CH2Cl2, -20 ˚C; (b) Li2CuCl4 (10 mol%), H2C=CHMgBr, THF, -60 to -40 ˚C; (c) TsCl, Et3N, DMAP, CH2Cl2; (d) KCN, 60% aq EtOH; (e) TMSCl, imidazole, DMF; (f) DIBAL-H, CH2Cl2, -78 to -45 ˚C.

Scheme 8

With cyclopropane 21 and aldehyde 30 in hand, we attempted the (3+2) annulation using the conditions that were effective during our model studies: 30 mol% tin(IV) chloride at 45 ˚C (Scheme  [8] ). Although cognizant that 30 could be prone to elimination in the presence of a Lewis acid, we were disappointed that this was indeed a serious complication. After reaction, no cycloadduct was detected by ¹H NMR spectroscopy, and the aldehyde had completely decomposed. Realizing that extensive optimization studies would be required, we decided to examine a model system to ease analysis.

First, a screen of Lewis acids was conducted with butylcyclopropane 31 and isovaleraldehyde (Scheme  [9] ). In previous studies, our laboratory had never extensively evaluated Lewis acids for use with alkylcyclopropanes and alkyl aldehydes. Ten Lewis acids (50 mol%) that were effective to varying degrees with aryl-aryl annulations were screened, but only aluminum(III) chloride and tin(II) triflate produced more than trace quantities of adduct 32 (entire screen not shown). These two Lewis acids were chosen for further study with the model aldehyde 33, which was synthesized from commercially available 4,4-dimethoxy-2-methylbutan-2-ol in two steps. The TBS ether was selected, because we reasoned its large size would inhibit elimination pathways by preventing coordination of the ether oxygen to a Lewis acid. We found that 33 decomposed in the presence of tin(II) triflate, but was only partially degraded by aluminum(III) chloride. Although aldehyde was present throughout the reaction, only the chloride 34 was isolated from the attempted annulation reaction. Ring opening by the hindered aldehyde was apparently unable to compete with chloride. [³¹]

Scheme 9 Initial results from alkylcyclopropane/alkyl aldehyde annulations

We also explored the use of less functionalized aldehydes that could be converted into ketones after annulation. Several asymmetric ketone allylations [³²-³4] are known that could ultimately be used to prepare the desired homoallylic alcohol; however, preliminary experiments with 35 [³5] and 36 [³6] were not encouraging. Because a route that uses these aldehydes is inherently less efficient but would still require optimization, we returned to TBS-protected hydroxy aldehyde 33.

Because we observed competitive chloride addition to 31, we hypothesized that using a non-nucleophilic anion might permit successful annulation. In this light, a series of aluminum bistriflimides were examined as potential catalysts (Table  [¹] ). These complexes were prepared in situ by protonolysis of trimethylaluminum with bis(tri­fluoromethane)sulfonimide. Aluminum(III) triflimide [Al(NTf2)3] was too active and decomposed 33, even at -70 ˚C. Reducing the Lewis acidity by replacing one bistriflimide with a methyl group resulted in a low yield of adduct 39. This product is a result of elimination of the TBSO group of the aldehyde prior to annulation. The first promising result was found using dimethylaluminum triflimide (Me2AlNTf2). While this catalyst still promoted elimination in both the aldehyde and cycloadduct, the desired adduct 37 made up 50% of the product mixture at 0 ˚C. This was increased to 66% at -20 ˚C, but further reduction to -35 ˚C hindered the annulation and instead catalyzed the trimerization of 33 to 40.

Yamamoto has reported the use of a diphenoxyaluminum bistriflimide for ketene cycloadditions. [³7] Using this precedent, we prepared several alkoxy- and phenoxyaluminum bistriflimide complexes. Although Yamamoto prepared the catalyst from trimethylaluminum and 2,6-diphenylphenol without isolation of the (ArO)2AlMe complex prior to protonolysis with bis(trifluoromethane)sulfonimide, we found that only catalysts directly prepared from isolated and recrystallized (RO)nAlMe3-n complexes were effective. The best catalysts were the hindered MABRNTf2 and MADNTf2 complexes (Table  [¹] ); both curtailed TBSO elimination in the aldehyde 33 and cycloadduct 37. MABRNTf2 gave the best diastereomeric ratio at 8.1:1, but 25 mol% MADNTf2 was used for further studies because of superior reaction efficiency.

The optimized conditions failed to facilitate the annulation of 21 and 33, affording only trace product after 10 hours at room temperature (Scheme  [¹0] ). We surmised that bicyclic cyclopropane 21 is less sterically accessible than butylcyclopropane 31, and hoped a reduction in the size of the protecting group would provide a solution. Because we had previously prepared aldehyde 30 with a TMSO group, we delved directly into the annulation of 21 and 30. Gratifyingly, the adduct 42 was formed, but in a modest 35-40% yield as a 10-11:1:1 mixture of inseparable diastereomers. The third diastereomer is formed by the combination of 21 and 30 with ent-30 and ent-21, respectively. MABNTf2, which contained a bulky triethylmethyl substituent, was prepared but did not increase the efficiency of the reaction.

Table 1 Screen of Aluminum Bistriflimidesa

entry Lewis acidb Temp (˚C) Yieldc (%) Ratio
37/38/39 d 37 (cis/trans)d
 1 Al(NTf2)3 -70  0 - -
 2 MeAl(NTf2)2   0 28 0:0:1 -
 3 -20 <5 0:0:1 -
 4 -35  0e - -
 5 Me2AlNTf2   0 63 2:1:1 5.5:1
 6 -20 72 4:1:1 5.5:1
 7 -35  0e,f - -
 8g t-BuOAlMeNTf2 r.t. 78 1.3:1:0.3 6.8:1
 9g DPPAlMeNTf2 r.t. 39 32:1:0 5.3:1
10g MABRNTf2 r.t. 55 1:0:0 8.1:1
11g MADNTf2 r.t. 71 85:1:0 6.1:1
12h r.t. 58 1:0:0 6.3:1

a All reactions were conducted with 3 equiv of 33 until complete consumption of 31, unless otherwise noted. The reaction times varied between 1.5 and 18 h.
b The aluminum bistriflimides were generated and used in situ by protonolysis of either AlMe3 or recrystallized (RO)nAlMe3-n complexes with HNTf2.
c Determined by ¹H NMR spectroscopy by integration relative to an internal standard.
d Determined by ¹H NMR spectroscopy.
e 0% conversion of 37.
f Compound 40 was isolated.
g 25 mol% catalyst used.
h 10 mol% catalyst used.

Allylic oxidation of 43 proved intractable. A variety of conditions were ineffective, including both selenium dioxide and transition-metal-catalyzed oxidations (Scheme  [¹¹] ). [³8-4¹] Oxidation of either the adduct 42 or allylcyclopropane 21 was also unsuccessful, forcing a change of strategy. We hoped that alkene isomerization to alkenyl-substituted tetrahydrofuran 44 would permit direct olefin oxidation, either through hydroboration or epoxidation. Several protocols have been developed that convert RCM catalysts into isomerization catalysts by the addition of diluted hydrogen or triethylsilane after metathesis is complete; [4²-44] however, with the exception of TMS deprotection, 43 was inert to these conditions and others using Crabtree’s or Wilkinson’s catalysts. We did not attempt these reactions with alternative protecting groups or after conducting the Krapcho decarboalkoxylation, primarily because we envisioned oxonene 44 could instead be readily synthesized from vinylcyclopropane 47, the use of which should also increase the efficiency of the (3+2) annulation.

The revised retrosynthesis, guided by the projected use of vinylcyclopropane 47, is similar to that presented for allylcyclopropane 21 (vide supra). The major changes are removal of a methylene spacer in the cyclopropane and the addition of a methylene group in the aldehyde (Scheme  [¹²] ). The transposition of the olefin in the oxonene 45 requires a direct olefin oxidation instead of an allylic oxidation.

Scheme 10 Annulation of 21 using optimized conditions

Scheme 11 Olefin metathesis and unsuccessful oxidation or isomerization

A racemic synthesis of 47 had already been accomplished during our model studies; however, we wished to pursue a more efficient synthesis that avoided protecting groups. From keto aldehyde 7, the (Z)-diene was installed using Yamamoto’s protocol in 71% yield (Scheme  [¹³] ). [45] Deprotonation with lithium 2,2,5,5-tetramethylpiperidide (LiTMP) and treatment with Mander’s reagent gave the keto ester 50; LiTMP was more effective than either LDA or LiHMDS. Diazo formation with p-ABSA proceeded cleanly, but various rhodium complexes provided poor yields of the desired vinylcyclopropane 47 (<39%). Cu(t-BuSal)2 (51) provided better selectivity for cycloprop­anation over C-H insertion, resulting in a 78% yield. [46]

Scheme 12 Retrosynthetic analysis with vinylcyclopropane 47

Scheme 13 Revised synthesis of vinylcyclopropane 47. Reaction conditions: (a) 1. (allyl)Ph2P, t-BuLi, THF, -78 to 0 ˚C, 2. Ti(Oi-Pr)4, -78 ˚C, 3. 7, -78 to 0 ˚C, 4. MeI, 0 ˚C to r.t.; (b) 1. LiTMP, THF, -78 ˚C, 2. HMPA; MeOC(O)CN; (c) p-ABSA, Et3N, MeCN; (d) 51 (4 mol%), benzene, reflux, slow addition of diazo over 20 h.

Heptenal 54 was synthesized by an analogous sequence to 30, with an allyl rather than vinylcuprate used to prepare the diol 52 (Scheme  [¹4] ). [³0] Sulfonylation, cyanide substitution, TMS protection, and DIBAL-H reduction all proceeded smoothly to produce aldehyde 54.

Scheme 14 Synthesis of heptenal 54. Reaction conditions: (a) Ti(Oi-Pr)4 (7.5 mol%), (-)-DET (10 mol%), TBHP, 4Å MS, CH2Cl2, -20 ˚C; (b) Li2CuCl4 (10 mol%), allylMgCl, THF, -60 to -20 ˚C; (c) TsCl, Et3N, DMAP, CH2Cl2; (d) KCN, 60% aq EtOH; (e) TMSCl, imidazole, DMF; (f) DIBAL-H, CH2Cl2, -78 to -45 ˚C.

As expected, the annulation of vinylcyclopropane 47 with 54 was higher yielding (76% vs. 41%), more diastereoselective (18:1 cis/trans), and occurred under milder conditions (10 mol% MADNTf2 at -30 ˚C) than the allylcyclopropane 21 (Scheme  [¹5] ); however, the olefin metathesis of adduct 55 failed under the conditions that were previously successful with adduct 42. Dimerization and macrocyclic dimerization to afford 56 and 57 were the only identifiable products of metathesis, though an appreciable amount of starting material was recovered. The striking contrast relative to the earlier metathesis with 42 is presumably a result of increased steric hindrance and reduced conformational flexibility of the C3 vinyl group.

Scheme 15 Annulation of vinylcyclopropane 47

After a brief survey of solvents, it was apparent that dichloromethane was optimal under high dilution (Table  [²] , entries 1-8). Concentrated solutions and aromatic solvents facilitated both dimerization and olefin isomerization/RCM, the latter resulting in the ring-contracted eight-membered oxonene 59. [47] The use of Grubbs’ 1st generation catalyst (G1) returned only starting material and uncyclized dimer 56 (entry 9). We found Grubbs’ 2nd generation (G2) and Hoveyda-Grubbs II catalysts (H-G2) were effective catalysts at elevated temperatures (60-80 ˚C) in sealed tubes, with H-G2 marginally more efficient. High catalyst loadings (20 mol% H-G2) and long reaction times (16-24 h) were required to force the reaction to completion and obtain a satisfactory 64% yield (entry 11). Unfortunately, the metathesis in a closed system is not reproducible upon ‘scale-up’ from 5 to just 20 mg (0.011 to 0.044 mmol), resulting in variable yields (35-62%). In search of practical conditions for scale-up, we were drawn to reports that sparging the reaction mixture with a stream of nitrogen can facilitate difficult RCM’s by actively removing ethene from the reaction. [48] [49] Although sparging at 40 ˚C in dichloromethane led to a sluggish reaction, switching to higher-boiling 1,2-dichloroethane yielded 69% of 58 in 1.3 hours using only 10 mol% H-G2 (entry 13). Identical results were obtained when conducted on a 0.36-mmol scale (70%). Experiments using 5 mol% of various catalysts, including the highly reactive Gre-II, [50] [5¹] led to diminished yields of 58 (entries 14-16).

Scheme 16 Ring-closing olefin metathesis and Krapcho decarboalkoxylation. Reaction conditions: (a) NaBr (10 equiv), aq DMF, 120 ˚C, 76% (90% conv.); (b) H-G2 (20 mol%), 0.0015 M, DCE, 80 ˚C, sealed tube; (c) H-G2 (10 mol%), 0.0011 M, DCE, 80 ˚C, N2 sparging; (d) NaBr (10 equiv), aq DMF, 120 ˚C; (e) BH3˙THF, Et2O-NMO, 4Å MS, CH2Cl2; TPAP; (f) 1. MePPh3Br (6 equiv), NaHMDS (5 equiv), toluene, 80 ˚C, 2. THF, 1.0 M HCl.

Table 2 Optimization of Ring-Closing Metathesis of 55

Entry Catalyst (mol%) Temp (˚C) Solvent Time (h) [55]0 (M) Yieldb (%)
55 58 59 57
 1 G2 (10) 90 toluene  3 0.011  0  0 27 34
 2 G2 (10) 75 benzene  2 0.0055 14 17 trace 29
 3 G2 (10) 80 benzene  3 0.0011 trace 11 27 20
 4 G2 (10) 40 CH2Cl2 14 0.055 trace  0 35 31
 5 G2 (10) 40 CH2Cl2 15 0.011 47  0 trace 15
 6 G2 (10) 40 CH2Cl2 15 0.0011 61 28  0  0
 7 G2 (10) 80 CH2Cl2  3 0.0011 12 55  0 trace
 8 G2 (10) 80 PhCF3  3 0.0011 20 40  0 11
 9 G1 (10) 40 CH2Cl2 15 0.011 48  0  0 trace
10 H-G2 (10) 80 CH2Cl2  5 0.0011 40 39  0  4
11 H-G2 (20) 80 CH2Cl2 24 0.0015  0 64  0  4
12 H-G2 (20) 80 DCE 15 0.0022 trace 39 11  8
13c H-G2 (10) 80 DCE  1.3 0.0016 trace 69  0 10
14c G2 (5) 80 DCE  5 0.0016 10 40  0  6
15c H-G2 (5) 80 DCE  5 0.0016 trace 55  0  8
16c Gre-II (5) 80 DCE  2 0.0016  0 62  0  9
a All reactions were conducted with 0.011 mmol 55 in a sealed tube at the indicated temperature unless otherwise noted.
b Determined by ¹H NMR by integration relative to an internal standard.
c Reaction conducted with slow sparging of the reaction mixture with nitrogen.

Krapcho decarboalkoxylation with sodium bromide in aqueous N,N-dimethylformamide at 120 ˚C provided 61 in 77% yield (Scheme  [¹6] ). A variety of salts were capable of promoting the reaction, but sodium bromide was tolerant of the labile TMS ether. Higher temperatures and the use of dimethyl sulfoxide led to unidentified byproduct formation. By reversing the order of the metathesis and decarboalkoxylation steps, only the macrocyclic dimer 60 was produced using the optimized closed system conditions: the ester is essential for a successful RCM. With the olefin 61 in hand, we attempted a hydroboration with borane-tetrahydrofuran complex followed by an oxidative workup. [5²] The ketone is also reduced during the hydroboration, but is restored in the presence of tetrapropylammonium perruthenate (TPAP) and N-methylmorpholine N-oxide. Fortunately, the reaction was modestly selective for the desired regioisomer; the dione 62 was produced in 49% yield. The use of (+)-diisopinocampheylborane [5³] completely controlled regioselectivity, but resulted in a reduced overall yield of 42%. The dione 62 was converted into the dienol 63 in excellent yield via a double methylene Wittig and treatment with aqueous hydrochloric acid in tetrahydrofuran.

Our attempts to perform an intramolecular etherification and olefin oxidation are summarized in Scheme  [¹7] . First, we sought to emulate the one-pot method developed by Kim via tandem intra- and intermolecular oxymercurations followed by reduction. Unfortunately, this sequence provided only ˜10% of 27 in a 6:1 mixture of diastereo­mers at C11. Attempts to perform the intramolecular oxymercuration with other mercury salts, such as Hg(OCOCF3)2 or Hg(OTf)2, were unsuccessful.

Boron trifluoride-diethyl ether complex [54] and bismuth(III) triflate [55] have both been used as hydroalkoxylation catalysts; however, their use only led to elimination and decomposition of 63. Iodoetherification of 63 with iodine/sodium hydrogen carbonate in acetonitrile resulted in 88% of the iodide 65. We envisioned a variety of oxidations could functionalize the remaining olefin, including epoxidation or halohydrin formation (Scheme  [¹7] , Conditions B). The former only resulted in decomposition, while N-bromosuccinimide gave several major products, with the bromohydrins formed unselectively. To enable epoxidation, the neopentyl iodide was reduced under forcing conditions [56] [LiEt3BH, THF-HMPA (4:1), 70 ˚C] to cleanly afford alkene 66. Epoxidation with dimethyldioxirane [57] provided a 1.7:1.0 mixture of epoxides as the sole products which, after reduction with lithium triethylborohydride, gave 64. Ultimately, the best route was found to be iodoetherification followed by oxymercuration and global reduction with tributylstannane. The dia­stereomeric ratio of 64 could be increased from 6:1 to 10:1 by repetitive flash chromatography. The ¹H and ¹³C NMR spectra were identical to that of deacetylpolyanthellin A (64).

Scheme 17 Synthesis of (+)-deacetylpolyanthellin A (64). Reaction conditions: (a) I2, NaHCO3, 4 Å MS, MeCN; (b) LiEt3BH, THF-HMPA (4:1), 70 ˚C; (c) Hg(OAc)2, acetone-H2O (1:1); (d) Bu3SnH, AIBN, benzene, 60 ˚C; (e) DMDO, acetone, 0 ˚C; (f) LiEt3BH, THF, 0 ˚C.

To complete the synthesis, 64 was acetylated (Ac2O, DMAP) to afford (+)-polyanthellin A (1) as a single dia­stereomer in 73% yield (Scheme  [¹8] ). The ¹H and ¹³C NMR spectra as well as the optical rotation was identical to that reported in the literature. Prof. Kim provided a synthetic sample for direct comparison.

Scheme 18 Completion of the synthesis of (+)-polyanthellin A (1)

Conclusion

In summary, (+)-polyanthellin A was synthesized in 15 linear steps from methallyl alcohol in 2% overall yield, in an average yield of 77% per step. The utility of the (3+2) annulation between cyclopropanes and aldehydes was demonstrated in a complex setting, wherein the reaction with sensitive protected β-hydroxy aldehydes was successful through the use of the potent but hindered Lewis acid catalyst, MADNTf2.

Purification of the reaction products was carried out either by acid/base extractive work-up or flash chromatography using Silia-P flash silica gel (40-63 µm) purchased from Silicycle. All reactions were carried out under an atmosphere of argon or N2 in oven-dried glassware with magnetic stirring. Yield refers to isolated yield of analytically pure material unless otherwise noted. Yields are reported for a specific experiment and as a result may differ slightly from those found in the tables, which are averages of at least two experiments. THF, Et2O, and CH2Cl2 were dried by passage through a column of neutral alumina under N2 prior to use. [58] Hexanes were dried by distillation from sodium metal immediately prior to use. HMPA, DMPU­, and Et3N were freshly distilled from CaH2. All other reagents were obtained from commercial sources and used without further purification unless otherwise noted.

Methyl (1 S ,3 R ,3a R ,7 R ,7a R )-7-Isopropyl-3-[( S )-2-methyl-2-(trimethylsilyloxy)hex-5-enyl]-4-oxo-1-vinyloctahydroisobenzofuran-3a-carboxylate (55)

A flame-dried vial equipped with a magnetic stirrer bar was charged with MAD (0.082 g, 0.17 mmol, 0.15 equiv) and anhyd CH2Cl2 (4.0 mL). Tf2NH (0.035 g, 0.11 mmol, 0.10 equiv) was added and the soln was stirred at r.t. for 10 min. A flame-dried 100-mL round-bottom flask equipped with a magnetic stirrer bar was charged with vinylcyclopropane 47 (0.270 g, 1.14 mmol, 1.0 equiv), aldehyde 54 (0.735 g, 3.43 mmol, 3.0 equiv), and anhyd CH2Cl2 (12.0 mL) and cooled to -78 ˚C. The catalyst soln was added and the reaction was stirred at -30 ˚C for 14 h. The soln was washed with sat. aq NaHCO3 (1 × 10 mL), dried (Na2SO4), and the solvents were removed under reduced pressure. The residue was purified by flash chromatography (linear gradient 2.5-5% EtOAc-hexanes) to leave 55 (0.388 g, 0.86 mmol, 76%) as a pale yellow oil; mixture of inseparable 2,9-cis/epi-C7/2,9-trans diastereomers 11:1:0.6; R f  = 0.42 (20% EtOAc-hexanes). Aldehyde 54 (0.391 g, 1.82 mmol, 71%) was also recovered.

[α]D ²4.9 +29.7 (c 1.02, CHCl3).

IR (thin film): 2958, 1717, 1463, 1434, 1250, 1042, 840, 755 cm-¹.

¹H NMR (400 MHz, CDCl3): δ = 5.92-5.73 (m, 2 H), 5.30 (d, J = 17.2 Hz, 1 H), 5.18 (d, J = 10.3 Hz, 1 H), 4.97 (dd, J = 17.1, 1.7 Hz, 1 H), 4.88 (d, J = 10.2 Hz, 1 H), 4.48 (dd, J = 9.3, 2.0 Hz, 1 H), 3.79-3.60 (m, 1 H), 3.72 (s, 3 H), 2.80 (t, J = 9.1 Hz, 1 H), 2.63 (ddd, J = 17.8, 6.6, 3.4 Hz, 1 H), 2.22 (ddd, J = 17.8, 10.9, 6.9 Hz, 1 H), 2.16-2.00 (m, 2 H), 1.87 (sext d, J = 6.7, 3.8 Hz, 1 H), 1.75 (ddd, J = 13.9, 10.5, 3.5 Hz, 1 H), 1.71-1.45 (m, 5 H), 1.29 (s, 3 H), 1.27-1.16 (m, 1 H), 0.92 (d, J = 6.8 Hz, 3 H), 6.82 (d, J = 0.81 Hz, 3 H), 0.08 (s, 9 H).

¹³C NMR (100 MHz, CDCl3): δ = 206.8, 170.8, 139.5, 137.4, 117.8, 113.7, 85.3, 77.7, 75.5, 67.6, 53.7, 52.5, 43.4, 42.8, 42.5, 37.7, 28.4, 27.8, 27.5, 21.5, 21.2, 15.9, 2.6.

LRMS (ESI): m/z [M + Na]+ calcd for C25H42NaO5Si: 473.3; found: 473.3.

Keto Ester 58

A flame-dried 1.0-L round-bottomed flask equipped with a magnetic stirrer bar and reflux condenser was charged with hydrobenzoisofuran 55 (0.163 g, 0.362 mmol, 1.0 equiv) and anhyd DCE (320 mL). H-G2 (0.023 g, 0.037 mmol, 0.10 equiv) was added. A stream of N2 was bubbled through the soln, which was stirred at 80 ˚C for 3.5 h, after which time the solvent was removed under reduced pressure, and the residue was purified by flash chromatography (linear gradient 5-10% EtOAc-hexanes) to give keto ester 58 (0.107 g, 0.253 mmol, 70%) as a clear, colorless oil; mixture of inseparable diastereomers 10:1; R f  = 0.28 (15% EtOAc-hexanes).

[α]D ²6.9 +98.9 (c 1.90, CHCl3).

IR (thin film): 2956, 1723, 1464, 1375, 1249, 1107, 1046, 925, 840, 754 cm-¹.

¹H NMR (400 MHz, CDCl3): δ = 5.51-5.30 (m, 2 H), 5.14 (dd, J = 13.1, 2.8 Hz, 1 H), 4.36 (br d, J = 5.6 Hz, 1 H), 3.71 (s, 3 H), 3.38 (dd, J = 8.8, 6.0 Hz, 1 H), 2.84 (td, J = 11.2, 8.8 Hz, 1 H), 2.53 (ddd, J = 17.5, 6.4, 3.8 Hz, 1 H), 2.34-2.21 (m, 2 H), 1.98-1.84 (m, 2 H), 1.84-1.58 (m, 3 H), 1.51 (t, J = 11.7 Hz, 1 H), 1.43 (s, 3 H), 1.28-1.18 (m, 1 H), 1.13 (dd, J = 14.3, 2.8 Hz, 1 H), 1.01 (d, 6.8 Hz, 3 H), 0.94 (d, 6.8 Hz, 3 H), 0.06 (s, 9 H).

¹³C NMR (100 MHz, CDCl3): δ = 204.2, 168.8, 131.7, 129.3, 85.1, 78.3, 78.0, 68.7, 53.2, 53.0, 45.9, 43.6, 42.9, 37.5, 29.0, 26.3, 22.9, 21.6, 20.8, 16.6, 2.8.

LRMS (ESI): m/z [M + Na]+ calcd for C23H38NaO5Si: 445.2; found: 445.2.

(+)-Deacetylpolyanthellin A (64)

A flame-dried vial equipped with a magnetic stirrer bar was charged with 63 (0.011 g, 0.036 mmol, 1.0 equiv), powdered 4Å MS (0.200 g), NaHCO3 (0.060 g, 0.72 mmol, 20 equiv), and anhyd MeCN (2.0 mL). The mixture was cooled to 0 ˚C, and I2 (0.018 g, 0.072 mmol, 2.0 equiv) was added at once. The mixture was left to stand in the refrigerator (˜4 ˚C) for 8 h. The mixture was diluted with sat. aq NaHCO3 (2 mL) and extracted with hexanes (3 × 3 mL). The combined organic fractions were washed with sat. aq NaCl (1 × 2 mL) and dried (Na2SO4). The solvent was removed under reduced pressure, and the residue was purified by flash chromatography (2.5% EtOAc-hexanes) to leave a mixture of products from decomposition on the silica gel. The residue was added to a vial equipped with a magnetic stirrer bar in acetone (1.0 mL) and deionized H2O (1.0 mL). Hg(OAc)2 (0.023 g, 0.072 mmol, 2.0 equiv) was added, and the mixture was stirred at r.t. for 2.5 h. Sat. aq NaCl (1.0 mL) was added and the mixture was stirred for 30 min. Then, the mixture was extracted with hexanes (3 × 4 mL) and dried (Na2SO4). This was filtered and evaporated to leave a white solid. This solid was dissolved in anhyd benzene (1.0 mL) after which Bu3SnH (0.042 g, 0.144 mmol, 4.0 equiv) and AIBN (0.0018 g, 0.011 mmol, 0.3 equiv) were added. The vial was purged with N2, sealed with a Teflon screw cap, and heated to 60 ˚C for 90 min. The solvents were removed under reduced pressure and the residue purified by flash chromatography (10% KF-silica gel, linear gradient 15-20% EtOAc-hexanes) to leave 64 (0.0065 g, 0.020 mmol, 56%) as a clear film; mixture of dia­stereomers 6:1. This material was repeatedly purified via flash chromatography (silica gel), collecting only the first fractions, to leave 64 (0.0034 g); mixture of diastereomers 10:1; R f  = 0.16 (20% EtOAc-hexanes).

IR (thin film): 3441, 2959, 2926, 1737, 1460, 1371, 1077, 1055, 956, 813 cm-¹.

¹H NMR (400 MHz, CDCl3): δ = 3.90-3.85 (m, 1 H), 3.58 (s, 1 H), 2.86 (td, J = 6.8, 1.7 Hz, 1 H), 2.42-2.28 (m, 1 H), 2.32 (dd, J = 11.1, 7.2 Hz, 1 H), 2.18 (dd, J = 14.2, 5.0 Hz, 1 H), 1.85 (d, J = 13.5 Hz, 1 H), 1.82-1.74 (m, 1 H), 1.68 (quint of d, J = 6.9, 2.1 Hz, 1 H), 1.57-1.32 (m, 8 H), 1.30 (s, 3 H), 1.30-1.23 (m, 1 H), 1.19 (s, 3 H), 1.17-1.08 (m, 1 H), 1.06 (s, 3 H), 0.93 (d, J = 7.0 Hz, 3 H), 0.83 (d, J = 6.9 Hz, 3 H).

¹³C NMR (100 MHz, CDCl3): δ = 93.3, 78.5, 75.5, 74.2, 70.3, 53.9, 47.7, 42.3, 41.8, 39.7, 36.6, 35.8, 33.8, 29.7, 29.6, 27.5, 21.8, 18.2, 17.5, 15.9.

LRMS (ESI): m/z [M + Na]+ calcd for C20H34NaO3: 345.23; found: 345.23.

(+)-Polyanthellin A (1)

A flame-dried vial equipped with a magnetic stirrer bar was charged with 64 (0.0034 g, 0.0105 mmol, 1.0 equiv), Et3N (0.073 mL, 0.053 g, 0.525 mmol, 50 equiv), DMAP (0.0016 g, 0.013 mmol, 1.25 equiv), and anhyd CH2Cl2 (2.0 mL). Ac2O (0.040 mL, 0.043 g, 0.42 mmol, 40 equiv) was added at once and the soln was stirred at r.t. for 1 h. The mixture was diluted with sat. aq NaHCO3 (2 mL) and extracted with Et2O (3 × 3 mL). The combined organic fractions were washed with sat. aq NaCl (1 × 2 mL) and dried (MgSO4). The solvent was removed under reduced pressure, and the residue was purified twice by flash chromatography (linear gradient, 5-7.5% EtOAc-hexanes) to leave 1 (0.0028 g, 0.0077 mmol, 73%) as a clear film; single diastereomer; R f  = 0.57 (25% EtOAc-hexanes). The spectral characteristics of the synthetic (+)-polyanthellin A (1) are in agreement with the natural product.

[α]D ²5.0 +9.9 (c 0.085, CHCl3) [Lit. [¹²a] [α]D ²0.0 +10.5 (c 0.31, CHCl3), Lit. [¹³] [α]D +8.9 (c 0.22, CHCl3), Lit. [¹4] ent-1 [α]D ²0.0 -9.9 (c 1.0, CHCl3).

IR (thin film): 2928, 2872, 1732, 1462, 1367, 1254, 1185, 1151, 1114, 1077, 1055, 1014, 991, 833, 800, 735 cm-¹.

¹H NMR (400 MHz, CDCl3): δ = 3.90 (td, J = 5.7, 1.1 Hz, 1 H), 3.54 (s, 1 H), 3.22 (td, J = 7.1, 2.2 Hz, 1 H), 2.46-2.38 (m, 1 H), 2.36 (dt, J = 13.1, 4.2 Hz, 1 H), 2.33-2.26 (m, 1 H), 2.19 (dd, J = 14.3, 4.9 Hz, 1 H), 2.00 (s, 3 H), 1.87 (dd, J = 14.1, 1.1 Hz, 1 H), 1.82-1.74 (m, 1 H), 1.70-1.60 (m, 1 H), 1.55-1.22 (m, 6 H), 1.48 (s, 3 H), 1.33 (s, 3 H), 1.22-1.12 (m, 2 H), 1.08 (s, 3 H), 0.93 (d, J = 6.9 Hz, 3 H), 0.80 (d, J = 6.8 Hz, 3 H).

¹³C NMR (125 MHz, CDCl3): δ = 170.3, 93.8, 83.2, 77.4, 75.5, 74.3, 51.0, 47.5, 42.3, 41.7, 39.7, 36.2, 35.6, 29.7, 29.6, 27.5, 24.1, 22.5, 21.7, 18.2, 17.5, 15.6.

HRMS (ESI): m/z [M + Cs]+ calcd for C22H36CsO4: 497.1668; found: 497.1681.

Acknowledgment

We thank Prof. Deukjoon Kim for providing an authentic synthetic sample of polyanthellin A. This research was supported by the NSF (CHE-0749691) and Novartis (Early Career Award to J.S.J.).

22

X-ray crystallography was performed by Dr. Peter White. CCDC 773310 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

22

X-ray crystallography was performed by Dr. Peter White. CCDC 773310 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

   Permissions and Reprints All articles of this category

Scheme 1 Approaches to the cladiellin core using oxocarbenium intermediates

Figure 1 Cladiellin diterpenes containing two ether bridges

Scheme 2 Tin(II) triflate catalyzed (3+2) annulation of 5 with benz­aldehyde

Scheme 3 Reaction conditions: (a) MVK, Et2NTMS (10 mol%), MeCN, reflux; (b) 1. PrPPh3Br, BuLi, THF, 0 ˚C, 2. 7; (c) 1. LDA, THF, -78 ˚C, 2. HMPA; EtOC(O)CN; (d) Mg(ClO4)2, I2, Et3N, CH2Cl2, r.t.; (e) 1. TBSO(CH2)3PPh3Br, BuLi, THF, 0 ˚C, 2. 7; (f) 1. LDA, THF, -78 ˚C, 2. HMPA; MeOC(O)CN; (g) p-ABSA, Et3N, MeCN; (h) Rh2(OAc)4 (3 mol%), CH2Cl2; (i) H2SiF6, MeCN; (j) 1. 2-O2NC6H4SeCN, Bu3P, THF, 2. H2O2.

Scheme 4 Reaction conditions: (a) PhCHO, Hf(OTf)4 (20 mol%), CH2Cl2, r.t., 4 h; (b) i-PrCHO, SnCl4 (30 mol%), DCE, 45 ˚C, 14 h; (c) i-PrCHO, SnCl4 (10 mol%), CH2Cl2, r.t., 1.5 h.

Scheme 5 Retrosynthetic analysis of polyanthellin A (1)

Scheme 6 Synthesis of allylcyclopropane 59. Reaction conditions: (a) MVK, 25 (5 mol%), 26 (20 mol%), neat, 4 ˚C; (b) 1. H2C=CH(CH2)2PPh3Br, BuLi, THF-DMPU (2:1), 2. 7, -78 to 0 ˚C; (c) 1. LDA, THF, -78 ˚C, 2. HMPA, 3. MeOC(O)CN; (d) p-ABSA, Et3N, MeCN; (e) Rh2(oct)4 (0.5 mol%), CH2Cl2.

Scheme 7 Synthesis of hexenal 30. Reaction conditions: (a) Ti(Oi-Pr)4 (7.5 mol%), (-)-DET (10 mol%), TBHP, 4 Å MS, CH2Cl2, -20 ˚C; (b) Li2CuCl4 (10 mol%), H2C=CHMgBr, THF, -60 to -40 ˚C; (c) TsCl, Et3N, DMAP, CH2Cl2; (d) KCN, 60% aq EtOH; (e) TMSCl, imidazole, DMF; (f) DIBAL-H, CH2Cl2, -78 to -45 ˚C.

Scheme 8

Scheme 9 Initial results from alkylcyclopropane/alkyl aldehyde annulations

Scheme 10 Annulation of 21 using optimized conditions

Scheme 11 Olefin metathesis and unsuccessful oxidation or isomerization

Scheme 12 Retrosynthetic analysis with vinylcyclopropane 47

Scheme 13 Revised synthesis of vinylcyclopropane 47. Reaction conditions: (a) 1. (allyl)Ph2P, t-BuLi, THF, -78 to 0 ˚C, 2. Ti(Oi-Pr)4, -78 ˚C, 3. 7, -78 to 0 ˚C, 4. MeI, 0 ˚C to r.t.; (b) 1. LiTMP, THF, -78 ˚C, 2. HMPA; MeOC(O)CN; (c) p-ABSA, Et3N, MeCN; (d) 51 (4 mol%), benzene, reflux, slow addition of diazo over 20 h.

Scheme 14 Synthesis of heptenal 54. Reaction conditions: (a) Ti(Oi-Pr)4 (7.5 mol%), (-)-DET (10 mol%), TBHP, 4Å MS, CH2Cl2, -20 ˚C; (b) Li2CuCl4 (10 mol%), allylMgCl, THF, -60 to -20 ˚C; (c) TsCl, Et3N, DMAP, CH2Cl2; (d) KCN, 60% aq EtOH; (e) TMSCl, imidazole, DMF; (f) DIBAL-H, CH2Cl2, -78 to -45 ˚C.

Scheme 15 Annulation of vinylcyclopropane 47

Scheme 16 Ring-closing olefin metathesis and Krapcho decarboalkoxylation. Reaction conditions: (a) NaBr (10 equiv), aq DMF, 120 ˚C, 76% (90% conv.); (b) H-G2 (20 mol%), 0.0015 M, DCE, 80 ˚C, sealed tube; (c) H-G2 (10 mol%), 0.0011 M, DCE, 80 ˚C, N2 sparging; (d) NaBr (10 equiv), aq DMF, 120 ˚C; (e) BH3˙THF, Et2O-NMO, 4Å MS, CH2Cl2; TPAP; (f) 1. MePPh3Br (6 equiv), NaHMDS (5 equiv), toluene, 80 ˚C, 2. THF, 1.0 M HCl.

Scheme 17 Synthesis of (+)-deacetylpolyanthellin A (64). Reaction conditions: (a) I2, NaHCO3, 4 Å MS, MeCN; (b) LiEt3BH, THF-HMPA (4:1), 70 ˚C; (c) Hg(OAc)2, acetone-H2O (1:1); (d) Bu3SnH, AIBN, benzene, 60 ˚C; (e) DMDO, acetone, 0 ˚C; (f) LiEt3BH, THF, 0 ˚C.

Scheme 18 Completion of the synthesis of (+)-polyanthellin A (1)