Asymmetric Formal Total Synthesis of the Stemofoline Alkaloids: The Evolution, Development and Application of a Catalytic Dipolar Cycloaddition Cascade

Abstract

A formal synthesis of didehydrostemofoline and isodidehydrostemofoline has been accomplished by preparing an intermediate in the Overman synthesis of these alkaloids from commercially available 2-deoxy-D-ribose. The work presented in this account chronicles the evolution of our explorations to identify the optimal steric and electronic control elements necessary to generate the tricyclic core structure of these alkaloids in a single operation from an acyclic precursor. The key step in the synthesis is a novel dipolar cycloaddition cascade sequence that is initiated by cyclization of a rhodium-derived carbene onto the nitrogen atom of a proximal imine group to generate an azomethine ylide that then undergoes spontaneous cyclization via dipolar cycloaddition. The synthesis features several other interesting reactions, including a Boord elimination to prepare a chiral allylic alcohol, a highly diastereoselective Hirama-Itô cyclization, and a useful modification of the Barton decarboxylation protocol.

1. Introduction

1.1. Isolation and Biological Activity

The Stemonacea family is a small group of flowering plants native to various regions of Southeast Asia.¹ Herbal extracts from a variety of these plants have been used for centuries as pesticidal agents and to treat respiratory diseases. These plant extracts have yielded a number of biologically active alkaloids that have been the focus of extensive biological and medical research.² Arguably the most complex members of the Stemona family of alkaloids are those of the stemofoline family, which are characterized by a caged hexacyclic architecture and differ in the geometry of the C(11)-C(12) double bond and the oxidation state of the four-carbon side chain R (Figure 1). Three species of the Stemona genus (S. tuberosa, S. japonica, and S. sessilifolia) are officially listed in the modern edition of the Pharmacopoeia of the People’s Republic of China as herbal antitussive agents, and the ground up roots of these plants are still sold in local markets and herb shops for medicinal and agricultural purposes.¹ Owing to the similar appearance of many of the Stemona species and their visual similarities to plants belonging to other genera, the incorrect common names are often used at these markets to sell plants that do not contain the active principles found in the Stemona species. Accordingly, one must be vigilant when studying or using the plant materials for medical or research applications. In fact, didehydrostemofoline (1) was first erroneously reported to be isolated from Asparagus racemosus and originally named asparagamine A.³ The roots of Asparagus racemosus, which are also sold as an herbal antitussive remedy, bear a striking resemblance to the roots of the Stemona plants, thereby giving rise to an early suspicion that 1 actually originated from a Stemona plant. Corroborating this hypothesis, 1 was later isolated from Stemona collinsae,⁴ and a recent report confirmed that 1 is not present in Asparagus racemosus altogether.⁵

Figure 1.

The Stemofoline Alkaloids

The stemofoline alkaloids were first isolated by Irie and coworkers from S. japonica,⁶ and they were later isolated from other Stemona species.⁷ Many of these alkaloids exhibit strong insecticidal activity because of their activity as insect acetylcholine (AChE) receptor antagonists.⁸ In a recent screen for AChE inhibitory activity, didehydrostemofoline (1) was found to be among the most potent of the stemofoline alkaloids.⁹ Didehydrostemofoline also exhibits in vivo anti-oxytocin activity and antitumor activity against gastric carcinoma,^4,10 whereas stemofoline (2) has been shown to be effective at increasing the sensitivity of clinically-used anticancer drugs such as paclitaxel, vinblastine, and doxorubicin by reversing P-glycoprotein mediated multi-drug resistance.¹¹ Continued interest in these alkaloids is reflected in more recent work in which a number of semisynthetic analogs were prepared and found to possess potent AChE inhibitory activity.^9,12

The complex molecular architecture coupled with the diverse biological activities of these alkaloids have inspired considerable interest from the synthetic community.^1,13,14 However, despite considerable effort, the only two accounts of the total syntheses of these alkaloids are Kende’s synthesis of (±)-isostemofoline (7)¹⁵ and Overman’s syntheses of (±)-1 and (±)-6.¹⁶ Other interesting approaches toward these alkaloids have also been reported.¹³ For example, Thomas applied an intramolecular Mannich reaction to construct the skeleton of stemofoline (2),^13e,d and Gin prepared the core structure of stemofoline by a novel process that featured an intramolecular dipolar cycloaddition.^13a,e

1.2. Total Syntheses of the Stemofoline Alkaloids

Both the Overman and Kende approaches to the stemofoline alkaloids relied upon the use of impressive cascade processes to construct the bridged polycyclic core (Scheme 1). In Kende’s synthesis of (±)-isostemofoline (7),¹⁵ the core structure was constructed at an advanced stage by a sequence that was initiated by treating the azabicycle 10 with TFA to effect MOM and BOC-deprotection and liberate the intermediate aminoalcohol 11, which spontaneously collapsed to the pentacyclic amine 12. The Overman approach¹⁶ to racemic 1 and 6 relied on the use of their prototypal aza-Cope-Mannich methodology in which 13 was heated with paraformaldehyde to generate the iminium ion 14, which underwent an aza-Cope reaction to give 15 that cyclized via an intramolecular Mannich reaction to deliver the tricycle 16.

Scheme 1.

Prior Approaches to the Stemofoline Core as Applied to Total Synthesis (R¹ = MOM, R² = TIPS)

Although the Kende and Overman syntheses set a high bar for the construction of the stemofoline core, the completed total syntheses were long requiring >30 steps. Furthermore, neither of these total syntheses was enantioselective. We thus believed that there was considerable opportunity to develop new chemistry that would lead to the first enantioselective syntheses of the Stemofoline alkaloids by a shorter sequence of reactions.

1.3. Initial Planning

Members of the Stemona alkaloids have long been a focus of attention in our group because of the many challenges associated with fabricating the polycyclic cores of these naturally-occurring bases. Our first introduction to these alkaloids resulted in an extraordinarily concise synthesis of croomine¹⁷ using sequential vinylogous Mannich reactions.¹⁸ We were similarly intrigued by the obvious challenges associated with developing short, enantioselective syntheses of representative members of the even more complex stemofoline group. Although the details of our plans to access these alkaloids have progressed through a number of iterations, the critical elements of our approach have remained the same (Scheme 2). In our original plan, we envisioned that the lactone 17 would serve as a key intermediate, and a number of tactics were envisioned for its conversion into didehydrostemofoline (1). A key step in producing 17 would involve the remote functionalization of the C(8)-position by a hydroxy radical 19 at C(2)¹⁹ that would be derived from the endo-alcohol 20 by reaction with hypervalent iodine as prescribed by seminal work from Suárez.²⁰ Alcohol 20 would be derived from the stereoselective reduction of ketone 21 by attack of a hydride reagent from the least hindered face of the ketone. The critical disconnection in the overall strategy, however, involved forming the tricyclic core of the stemofoline alkaloids by the intramolecular 1,3-dipolar cycloaddition of an olefinic azomethine ylide of the general form 22. The key challenge was to identify an azomethine ylide 22 that was suitably substituted so it would undergo a highly regioselective dipolar cycloaddition to give 21 (Scheme 2). However, the path to identifying the optimal azomethine ylide 22 and its precursor 23 was fraught with unanticipated pitfalls. It was necessary to prepare and investigate several functionalized pyrrolidine derivatives 23 having various alkyl and heteroatom substituents at R² and R³ and a number of different Z groups. The precise nature of the azomethine ylide precursor 23 therefore underwent a series of iterations that were guided by an evolving understanding of this cycloaddition.

Scheme 2.

Novel 1,3-Dipolar Cycloaddition Approach to Stemofoline Core

1.4. First Generation Approach

We first envisioned that the azomethine ylide 22 might be generated by expelling the leaving group Z from pyrrolidinone 23, followed by deprotonation of the resulting iminium ion (Scheme 2). In order to test the feasibility of this plan, we used pyrrolidinone 24 as a model to screen conditions for azomethine ylide formation (Scheme 3). However, our efforts were quickly derailed when we discovered that all attempts to replace the Boc-protecting group of 24 with a variety of functionalized methylene groups gave the pyrrole 26 as the only identifiable product.

Scheme 3.

Initial Attempts to Form Precursors of Azomethine Ylides

We reasoned that the enol form of 25 (R = H) underwent rapid and unavoidable oxidation, so we decided to reduce the ketone moiety in 24 to obviate any possibility of enolization during the process of refunctionalizing the nitrogen atom. In the event, 24 was reduced, and the requisite α-amino nitrile moiety was installed by removing the Boc-protecting group and subjecting the intermediate secondary amine to a Strecker reaction to provide 27 (Scheme 4). As we had predicted, the amino nitrile 27 did not undergo aromatization to a pyrrole, but all attempts to convert 27 into the corresponding azomethine ylide 28 by the net loss of HCN were unsuccessful.

Scheme 4.

Discovery of Oxidative Azomethine Ylide Formation

Generation of an azomethine ylide 28 from 27 requires ionization by loss of cyanide ion and the proton at C(3) (Scheme 4). We thus queried whether azomethine ylide generation might be facilitated by increasing the acidity of this proton by oxidation of the alcohol moiety. Accordingly, 27 was oxidized by a Swern oxidation with careful exclusion of oxygen to avoid pyrrole formation. Much to our surprise, we isolated a mixture (ca. 5:1) of cycloadducts 33 and 32 in 69% combined yield; both of these cycloadducts retained cyano group.^14a,21 We believe this unusual transformation involved formation of the cyano-azomethine ylide 29 that then underwent cycloaddition by the two regioisomeric transition states 30 and 31 to furnish 32 and 33, respectively. Although this reaction sequence provided compelling support for the viability of our approach to the stemofoline core, we were unable to decyanate 33 to provide 34. We were thus obliged to modify the strategy in order to access a tricyclic intermediate without the superfluous functionality at C(5).

1.5. Second Generation Approach

Our second approach (Scheme 5) was inspired by work reported by Joucla,²² who had shown that flash vacuum thermolysis of oxazolidines generated azomethine ylides that underwent dipolar cycloadditions with olefinic partners. We thus discovered that heating the oxazolidine 35a in a sealed tube gave a mixture (1:3) of the regioisomeric cycloadducts 39a and 40a in 96% combined yield,²¹ presumably via the corresponding transition states 37a and 38a (Scheme 5). Although the yield of this transformation was high, the undesired regioisomer 40a was favored. As a possible tactic to enhance formation of the desired regiochemistry in the cycloaddition, we explored the possibility that the undesired transition state 38b might be destabilized by the presence of a bulky substituent R at C(9). In order to test this hypothesis, 35b was prepared and subjected to thermolysis to provide a slightly more favorable mixture (1:1) of cycloadducts 39b and 40b in 95% yield. ²¹

Scheme 5.

Effect of Steric Factors on Dipolar Cycloaddition

The results of the cycloadditions of the azomethine ylides 36a and 36b validated our hypothesis that steric effects can be exploited to favorably affect the regiochemical course of the cycloaddition. The modest increase in selectivity that was observed, however, was insufficient for the purpose of a total synthesis. Accordingly, we continued our search to identify what structural modifications to the azomethine ylide precursor were required to further enhance the selectivity.

2. Results and Discussion

2.1. Development of a New Synthetic Plan

Our early work established the underlying viability of the key cycloaddition strategy for constructing the core of the stemofoline alkaloids.^14a,b It also provided some direction as to how steric and electronic parameters influenced the regioselectivity of the cycloaddition. An important goal in refining our approach was to enable an enantioselective synthesis of the tricyclic core utilizing a starting material from the “chiral pool” as the sole source of chirality. Remaining stereocenters would be introduced by diastereoselective transformations that would be subject to substrate-controlled reactions. We reasoned that such a strategy would likely increase the overall efficiency of the synthesis by decreasing the number of synthetic operations and the need for using multiple stoichiometric chiral auxiliaries that characterized our initial efforts.

Toward this end, we viewed commercially available 2-deoxy-D-ribose (47) as a suitable starting material because it possesses the resident chirality, and it contains five of the nine carbon atoms found in the tricyclic core of the stemofoline alkaloids. The essential features of our new strategy are outlined in retrosynthetic format in Scheme 6. The first stage of the synthesis would thus involve converting 2-deoxy-D-ribose into the allylic alcohol 46 utilizing sequential Wittig olefination and Boord elimination reactions. The chirality of the allylic alcohol at C(8) would then be exploited to stereoselectively establish the amino functionality at C(9a) to provide 45 via a diastereoselective, intramolecular aza-Michael reaction. Elaboration of 45 via a Claisen condensation followed by diazo transfer and refunctionalization of the aminoalcohol leads to 44, the immediate precursor of the azomethine ylide 43. The critical question was whether the presence of a protected hydroxyl group at C(8) would disfavor the transition state 48 leading to the undesired regioisomer and preferentially deliver 41 via transition state 42.

Scheme 6.

Revised Retrosynthesis of Didehydrostemofoline

2.2. 2-Deoxy-D-Ribose as a Chiral Starting Material for Allylic Alcohol 42

There was some precedent for preparing the allylic alcohol 46 from 2-deoxy-D-ribose (47),²³ but 46 had only been isolated as a byproduct of another process. A reliable means for preparing 46 had thus not appeared in the prior art. Accordingly, we sought to develop a concise and effective process to synthesize 46. The first generation route to 46 commenced with protecting 2-deoxy-D-ribose (47) by treatment with 2-methoxypropene in the presence of PPTS to give acetonide 49 (Scheme 7).²⁴ In order to obtain reproducible yields, it was necessary to perform the reaction at a concentration of 0.3 M rather than at 0.8 M as described in the literature; however, the product was always accompanied by the formation (>20%) of the protected furanoside (5-membered, α and β) forms of the sugar, which were separated by chromatography. The protected hemiacetal 49 was then subjected to a Wittig olefination to provide enoate 51 as an E:Z-mixture (5:1) of olefin isomers in 96% yield. We had previously shown that use of a tri-n-butylphosphine derived Wittig reagent enhanced the diastereoselectivity in related olefinations,²⁵ and application of this modification to 49 afforded a markedly improved E:Z-ratio of ~9:1. The disadvantage of this procedure is that the requisite Wittig reagent is more expensive.

Scheme 7.

Initial Synthetic Efforts to Prepare 46

It was also known that molecular iodine can catalyze the thermodynamic equilibrations of enones and enoates.²⁶ Because the next step in the sequence would involve the use of iodine to convert the primary alcohol into an iodide, we queried whether the presence of an excess of iodine might effect concomitant epimerization of the undesired Z-enoate. In the event, reaction of a E:Z-mixture (5:1) of 51 with triphenylphosphine and a 10 mol % excess of iodine provided exclusively the E-iodoenoate 52 in 73% yield. Upon treatment with activated Zn (dust)/AcOH, compound 52 underwent facile Boord elimination²⁷ to give a mixture (1:2) of the desired allylic alcohol 46 and the cyclopentane 53 in a combined 75% yield. Although samarium iodide mediated radical cyclizations of substrates like 52 to give cyclopentanes such as 53 are known,²³ the reaction of zinc metal with 52 was expected to give predominately the elimination product 46 rather than 53.

This route to the alcohol 46 thus proceeded in 10% overall yield from 2-deoxy-D-ribose by a sequence that required four synthetic steps and four chromatographic purifications. However, there are some obvious deficiencies, and we set to the task of developing a more efficient process. Toward this end, we reasoned that the yield of the Boord elimination might be improved by refunctionalizing the diol to increase the leaving group ability of the secondary alcohol at C(7), thereby favoring elimination over cyclization (Scheme 7). Acetonide 52 was thus converted into iodo diacetate 54 in 96% yield by sequential treatment with Dowex/MeOH and acetic anhydride in the presence of 4-dimethylaminopyridine (DMAP). When the iodide 54 was treated with Zn (dust) in methanol, the ratio of elimination to cyclization improved, and the alcohol 46, which was formed by transesterification under the reaction conditions, was isolated in 46% yield. This alternate route provided 46 in a slightly improved 16% overall yield, but six synthetic operations and four chromatographic purifications were now required. However, a key discovery was that the Boord elimination could be improved by enhancing the leaving group ability of the oxygen function at C(7).

In the interest of further streamlining the synthesis of 46, the diol protection strategy was more closely scrutinized. Because unprotected 2-deoxy-D-ribose (47) was known to undergo olefination,²⁸ it was first allowed to react with Wittig reagent 50, and the resulting mixture of enoates was treated directly with Ph₃P/I₂ to effect iodination (Scheme 8). The reaction was subjected to an aqueous workup, and the crude mixture of iodo diols was acetylated to provide the iodo diacetate 54 in 87% over two steps. Although 54 could be readily purified by chromatography, simple filtration through a plug of silica gel provided material that was sufficiently pure for the next step. After additional experimentation to optimize the Boord elimination, we discovered that the yield of product was improved significantly by the use of Zn granules rather than Zn dust. Under these modified conditions, the allylic alcohol 46 was formed in 62% yield.

Scheme 8.

Optimized Synthesis of 46 (3 Steps, 1 Column, 54% Overall Yield)

This result is somewhat surprising because there is no apparent reason that the physical state of the Zn metal should so drastically affect the product distribution. On the other hand, the presence of trace metal impurities in the Zn dust might catalyze the conjugate addition, thereby placing importance on the source and purity of the zinc metal. However, the assay on the batches of Zn dust and Zn granules that were used did not reveal any differences in impurity profiles. This uncertainty notwithstanding, it was possible to perform this reaction reproducibly on scales up to 100 g. We were now able to prepare 46 in 54% overall yield by a process that required three chemical operations and only one chromatographic purification.

2.3. Application of Hirama-Itô Cyclization

With an efficient and scalable route to the allylic alcohol 46, we were positioned to probe the efficacy of the Hirama-Itô cyclization,²⁹ which involves an intramolecular Michael addition of O-tethered carbamates to α,β-unsaturated carbonyl compounds. The allylic alcohol 46 was first treated with chlorosulfonylisocyanate to give the primary carbamate 55 in nearly quantitative yield after a hydrolytic workup (Scheme 9). Reaction of 55 with NaH in THF at room temperature gave the cyclic carbamate 45 as a mixture (4:1) of syn- and anti-diastereomers in 75% yield via the preferred transition state 56 in which the vinyl group resided in an equatorial orientation. This experiment clearly established that the stereochemistry at C(8) could be exploited in a substrate controlled reaction to create the stereocenter at C(9a). Despite this swift success, it was apparent that further experimentation would be needed to improve the diastereoselectivity of this intramolecular aza-Michael reaction. Toward this goal, the work of Hirama and Itô offered little encouragement because results of those studies suggested that varying reaction parameters would not likely have a significant impact upon the yield and diastereoselectivities of the process. Undaunted, we discovered that a number of variables impacted the yield and diastereoselectivity of this specific conversion.

Scheme 9.

Application of the Hirama-Itô Cyclization Reaction

Although not reported by Hirama and Itô, the elimination product 57 often accounted for 10–20% of the product mixture upon cyclization of 55 (Scheme 9). We discovered that the extent of this elimination increased when more polar solvents were employed, and the use of DMF provided as much as 20% of 57. The use of NaH in THF did not provide appreciable conversion at temperatures of 0 °C or below, but when CH₂Cl₂ was used as the solvent, the reaction was complete within two hours at −10 °C with only trace amounts of 57 being formed. We examined the effects of changing the cationic counterion and found that more Lewis acidic counterions led to increasing amounts of elimination, but the Lewis acidities had no influence on the diastereoselectivity. For example, the use of KOt-Bu in THF at −20 °C gave nearly 20% of 57; however, the use of LiOt-Bu under the same conditions provided the triene 57 almost exclusively. Varying the temperature of the reaction had little effect on the amount of elimination, but the diastereoselectivity was sensitive to changes in temperature. For example, at temperatures below −20 °C, the use of freshly sublimed KOt-Bu in THF gave a mixture (10:1) of syn- and anti-45 in 78% yield. When the reaction was performed at −10 °C using NaH as the base in CH₂Cl₂, a slightly less favorable mixture (8:1) of syn- and anti-45 was obtained in 80% yield. However, under these conditions the reaction mixture was contaminated with fewer impurities, and the desired 45 could be readily isolated by a single recrystallization as an inseparable mixture (8:1) of diastereomers. Accordingly, these conditions were adopted as the standard for preparing 45.

2.4. Elaboration of 45 to Precursor of the Cycloaddition

The next stage of the synthesis involved converting 45 into an orthogonally protected amino alcohol of the general form 44 (Scheme 6) that bore the required diazo-β-ketoester moiety, and several approaches were examined. Carbamates such as 45 are known to be difficult to cleave directly, but the corresponding N-acyl derivatives can be readily opened by nucleophilic attack. Accordingly, the carbamate moiety in 45 was activated by N-acylation, and the syn Boc-carbamate 58 thus obtained underwent facile methanolysis using Cs₂CO₃/MeOH to deliver the amino alcohol 59 (Scheme 10). Although 59 could be isolated in relatively pure form, we found that it cyclized readily to give the lactone 60 on attempted purification and thus decided to test 60 as an intermediate. Toward this end, it was necessary to optimize the conversion of 59 into 60, and we found that this was most conveniently achieved by adsorbing crude 59 onto dry silica gel, followed by column chromatography to provide lactone 60 in 80% yield from 58.

Scheme 10.

Synthesis of Diazoacetoacetate 63 from Cyclic Carbamate 45

The cross Claisen reaction of lactone 60 with the enolate of methyl acetate gave the hemiacetal 61, but the yield was invariably low. We examined a number of experimental variants, including the use of tert-butyl acetate to minimize the amount of homo-Claisen product formed, but the yield of 61 could not be improved. We also found that performing the Claisen reaction on crude 59 provided 61 directly, albeit also in modest yield. That the lactone 60 was a likely intermediate in this transformation is supported by the independent observation that treatment of 59 with excess LDA gave 60. In practice, using 59 as the substrate for the Claisen reaction was a more convenient process for preparing 61. We then found that 61, which is presumably in equilibrium with its acyclic form, could be readily converted into 62 by diazo transfer using p-acetamidobenzenesulfonyl azide (p-ABSA) to give the diazo compound 62 in three steps and 29% overall yield from 58. Subsequent protection of the hydroxyl group as its TBDPS-ether proceeded in 79% yield to provide the key diazoacetoacetate intermediate 63.

This route to 63 did provide small quantities of material for exploring subsequent reactions, but the conversion of 60 into 63 was sufficiently laborious that we decided to develop a more expedient route that avoided the intermediate hemiacetal 61. We found after some experimentation that crude 59 underwent facile silylation using TBDPS-Cl (1.5 equiv) and imidazole (1.2 equiv) to give 64 in 80% yield from 59 (Scheme 11). The yields were consistently higher when the TBDPS-Cl was premixed with imidazole in DMF for 15–20 min prior to addition of 59. If imidazole was used in excess of the silyl chloride, only the lactone 60 was isolated.

Scheme 11.

Optimized Claisen Condensation

When 64 was subjected the Claisen reaction conditions that were originally developed to prepare 61, the β-ketoester 65 was isolated in 60% yield (Scheme 11). However, loss of the N-Boc protecting group from product 65 occurred as a significant side reaction, providing 20–30% of the free amine as a side product. After performing several control experiments, we determined that the N-Boc group in 65 was not stable in the presence of excess LDA. Fortunately, we also discovered that use of NaHMDS as a base was not plagued by this side reaction, and 64 was thus transformed into the desired β-ketoester 6 in 75% yield;5 17% of 64 was also recovered. The reaction of 65 with p-ABSA then furnished the requisite diazoacetoacetate 63 in 92% yield. The synthesis of 63 from commercially available 2-deoxy-D-ribose was thus accomplished in 10 synthetic steps and in 21% overall yield. This process for preparing 63, which has been reproduced multiple times on a decagram scale, requires only four chromatographic purifications.

2.5. Development of a Regioselective Cycloaddition

With significant quantities of the diazoacetoacetate 63 in hand, the stage was set to examine the pivotal dipolar cycloaddition step. Toward this end, the diazo compound 63 was first converted into pyrrolidinone 66 in 86% yield by a rhodium catalyzed NH-insertion according to a protocol we had previously utilized (Scheme 12).¹⁴ When 66 was allowed to react with dimethoxymethane in the presence of trifluoroacetic acid (TFA), the oxazolidine 67 was formed in 33% yield. We tentatively attributed the low yield in this transformation to loss of the silyl protecting group and reasoned that this problem could be eventually be solved with a different protecting group. In the end, this was not necessary because thermolysis of 67 delivered an unfavorable mixture (1:2) of regioisomers 71 and 72. This discouraging experiment revealed that the protected hydroxyl group at C(8) was not as effective in directing the regiochemical outcome of the dipolar cycloaddition as was a substituent at C(9) (cf 35b → 39b + 40b, Scheme 5).

Scheme 12.

Oxazolidine Thermolysis of a C(8)-Substituted Precursor

At this juncture, it was apparent that the mere presence of substituents at C(8) and C(9) was not sufficient to guarantee high regioselectivity in the dipolar cycloaddition. Indeed, the only azomethine ylide that underwent a cycloaddition with a significant level of regioselectivity had an electron withdrawing group at C(5) (cf 29). We therefore turned our attention to generating an azomethine ylide bearing a readily removable, electron-withdrawing group at this position. After considering several options, we decided to examine the approach outlined in Scheme 13. This unusual sequence was inspired by the knowledge that stabilized metal carbenes can react with imines via intermolecular processes to generate azomethine ylides.³⁰ The reactions of metal carbenes to form azomethine ylides in an intramolecular sense, however, are not well known.³¹ Moreover, the formation of an azomethine ylide by such a cyclization has only been studied in intermolecular dipolar cycloaddition reactions with highly active dipolarophiles.³¹ In a considerably more ambitious adaptation of this reaction, we envisioned that the reaction of diazoacetoacetate 73 with a rhodium catalyst would initiate a novel cascade of reactions in which the initially formed metal carbene would cyclize onto the nitrogen atom of the proximal imine group to generate the azomethine ylide 74 that would then undergo spontaneous cyclization via a dipolar cycloaddition to deliver the desired adduct 76.

Scheme 13.

Dipolar Cycloaddition Cascade Approach to the Stemofoline Core

As the first step toward implementing this new plan, carbamate 63 was converted into diazoimine 73 by removing the Boc-group and treating the intermediate ammonium salt with NEt₃ and benzyl glyoxylate to provide the diazoimine 73 in quantitative yield (Scheme 13). Anxious to test the feasibility of the key cascade sequence, crude 73, which contained an equimolar amount of TFA•NEt₃, was used in initial experiments. As will soon be appreciated, this was a fortunate decision with unanticipated consequences. In the event, the crude diazoimine 73 thus obtained was heated in refluxing benzene in the presence of Rh₂(OAc)₄ (5 mol %) to give the tricycle 76 in 35% yield; none of the regioisomeric cycloadduct was isolated, although a mixture of a number of other side products was obtained. Based upon mechanistic considerations and the ¹H NMR spectrum of the mixture, we surmised that at least some of these compounds were isomeric aziridines. Because aziridines are known to undergo thermal ring opening to generate azomethine ylides,³² we conducted the reaction at higher temperature in refluxing xylenes and obtained 70 in 75% yield. In this case, the reaction mixture was considerably cleaner than at lower temperature, and the putative isomeric aziridines were no longer observed in the reaction mixture. The catalyst loading could be lowered to 3 mol % without deterioration in yield. This remarkably efficient cascade reaction sequence is notable because the tricyclic core of the stemofoline alkaloids is generated with high stereoselectivity and regioselectivity in a single chemical operation from an acyclic precursor.

Toward improving the yield of 76 from 63, the intermediate diazoimine 73 was purified by column chromatography on basic alumina. Surprisingly, however, when the pure sample of 73 was heated in refluxing xylenes in the presence of Rh₂(OAc)₄, a mixture (1.5:1) of cycloadducts 76 and 82 was obtained in 66% combined yield (Scheme 14).

Scheme 14.

Mechanistic Rationale for Observed Cycloaddition Selectivity

The divergent results obtained with crude and purified 73 beg an explanation (Scheme 14). One reasonable possibility is that the rhodium carbene that is generated from 73, which presumably has the imine stereochemistry shown in 77, undergoes kinetic cyclization to form the U-shaped azomethine ylide 78. This intermediate may either undergo dipolar cycloaddition via the cis-regioisomeric transition states 79 or 80 or isomerize to the S-shaped ylide 74, which should be more stable owing to reduced steric A^1,3-interactions. The observed effect of the presence of NEt₃•TFA during the sequence suggests that the isomerization of 78 to 74 may be acid-catalyzed. The S-shaped ylide 74 may undergo cyclization via the two competing, regioisomeric transition states 75 and 81 to give 76 and 84, respectively; whereas 78 can lead to 82 and 83 via the corresponding transition states 79 and 80. Adducts 83 and 84 were not observed, presumably because of the unfavorable interactions between the two ester groups in transition state 80 and the ester and the protected alcohol in transition state 81. The cycloadduct 82 was only isolated if the reaction was conducted in the absence of NEt₃•TFA. It thus tentatively appears that the isomerization of 78 to 74 in the presence of acid is more facile than the cyclization of 78 to give 82.

2.6. Removing the Now Superfluous Ester Group at C(5)

Inasmuch as the ester group had now served its critical role of guiding the regiochemistry of the dipolar cycloaddition reaction, it was time to examine tactics to effect its removal from 76. Toward the dual objectives of removing the ester at C(5) and oxidizing C(8) by remote functionalization, the keto function in 76 was reduced with NaBH₄ in methanol at −30 °C to provide the endocyclic alcohol 85 in 86% yield as the only diastereomer. When 85 was subjected to the Suárez protocol for remote radical functionalization,²⁰ the tetracycle 86 was obtained in 94% yield. Hydrogenolysis of the benzyl ester moiety in 86 provided the amino acid 87 in virtually quantitative yield. Williams and coworkers had recently reported that chloroform could be used as a hydrogen atom donor in developing a one-step, Barton decarboxylation process that seemed nicely suited to the task at hand.³³ Indeed, after forming the Barton ester derived from amino acid 87 using N-hydroxy-2-pyridinethione (88) and dicyclohexylcarbodiimide (DCC) in CHCl₃, the reaction mixture was irradiated with a tungsten filament light bulb (250 W) for ~1 h at room temperature to furnish the decarboxylated tricycle 89 in 40% yield. The major byproduct obtained was pyridyl sulfide 90, which was also observed in the absence of a suitable hydrogen donor by Barton.³⁴ Toward improving the yield of 89, more reactive H-atom donors were examined as additives, and we eventually discovered that use of tert-BuSH (10 equiv) increased the yield of 89 to 71%.

2.7. Completion of Formal Total Synthesis

With the advanced intermediate 89 in hand, a number of endgames were contemplated. One attractive strategy featured the formation of the lactone ring in 90 via a dirhodium catalyzed, regioselective C–H insertion to directly functionalize the C(9)-position of the stemofoline core (Scheme 16). Such reactions are known to favor insertion into 3° C–H bonds over 2° C–H bonds, but it is possible to use chiral catalysts in simpler, cyclohexanol-derived systems to override this innate preference.³⁵ Because C-H insertions of this type had never been applied to a substrate as complex as 91, there was a significant opportunity to advance this methodology.

Scheme 16.

Retrosynthetic Analysis Inspired by Regioselective C–H Insertion

In order to assess the viability of the endgame strategy outlined in Scheme 16, the ester function at C(3) of 89 was reduced with DIBAL-H to provide aldehyde 92 in 81% yield (Scheme 17). When 92 was subjected to a Julia-Kocienski olefination with tetrazole 93 in the presence of KHMDS,³⁶ the crude olefin 94 that was obtained was directly subjected to desilylation with tetra-n-butylammonium fluoride (TBAF) to afford hemiketal 95 in 94% yield over the two step sequence. Initial attempts to acylate the hemiketal hydroxy group to directly generate the requisite diazoacetate 91 involved use of p-toluenesulfonylhydrazone of glyoxylic acid chloride according to procedures reported by House³⁷ and Corey.³⁸ We were attracted to this route because we had employed this reagent with considerable success in the past.³⁹ However, despite repeated attempts, we were not able to access the diazoacetate 91 from hemiketal 95. Fukuyama had reported an alternative method for generating diazoacetates from bromoacetates,⁴⁰ so we turned our attention to converting 95 into 96. Unfortunately, use of a number of reagent combinations (i.e., bromoacetylbromide/base, bromoacetylbromide/HBr, and bromoacetic anhydride/DMAP) all failed to give significant quantities of 96.

Scheme 17.

Advancement of Decarboxylated Core

We reasoned that the hemiketal nature of the tertiary-like hydroxyl group at C(8) might be adversely affecting its reactivity, so we decided to switch to a modified substrate in which the hydroxyl group at C(8) was a simple secondary alcohol. In the event, protection of the tricyclic alcohol 85 as its MOM ether furnished 97 in 75% yield (Scheme 18). Hydrogenolysis of the benzyl ester in 97 afforded the free carboxylic acid, which was then subjected to the same modified Barton decarboxylation protocol used previously (see Scheme 15, 87→89) to give 98 in 63% overall yield from 97. Reduction of the ester group in 98 followed by a Julia-Kocieński olefination gave 100 in 80% yield from 98. Desilylation of the TBDPS-ether 100 afforded alcohol 101 in 95% yield. After screening several reaction conditions to acylate the C(8) alcohol of 101, we found that bromoacetate 102 was best prepared using the esterification conditions developed by Steglich (bromoacetic acid, DCC, DMAP).⁴¹ Bromoacetate 102 was then subjected to conditions developed by Fukuyama (TsNHNHTs, DBU)⁴⁰ to prepare diazoacetate 103.

Scheme 18.

Attempted C-H Insertion

Scheme 15.

Advancement of Cycloadduct and Decarboxylation of Core

Although this conversion had been reported to work well with a number of simple compounds,⁴⁰ we discovered that the reaction of 102 to give 103 was highly problematic, and the major product was invariably 101. The diazoacetate 103 was also surprisingly unstable, and all attempts to purify it led to significant losses of material. Consequently, crude 103 that had simply been filtered through a short pad of neutralized silica gel was used in attempts to functionalize the C(9)-position of 103 by C–H insertion, and we were able to screen several rhodium catalysts using this material. However, exposure of 103 to Rh₂(OAc)₄, Rh₂(5(S)-MEPY)₄, and Rh₂(5(R)-MEPY)₄ did not provide any detectable quantity of the desired lactone 104. Given the difficulties we encountered in trying to prepare the sufficient amounts of diazo compound 103 for C–H insertion experiments, we turned our attention to an alternate approach in which we would convert 101 into an intermediate in Overman’s syntheses of racemic stemofoline alkaloids.¹⁶

We thus subjected 101 to a Parikh-Doering oxidation to afford ketone 105,⁴² the enolate of which was alkylated with ICH₂CO₂Et to provide the axial-alkylated product 106. Epimerization of 106 under basic conditions followed by treatment with TFA delivered the hemiketal 107 in 81% yield over the two steps. Because racemic 107 has been converted into didehydrostemofoline (1) and isodidehydrostemofoline (6) by Overman, its preparation completes the formal enantioselective syntheses of these alkaloids. The spectral data of our synthetic (+)-107 is consistent with those reported by Overman for (±)-107.¹⁶

3. Conclusion

In summary, a formal synthesis of didehydrostemofoline (1) and isodidehydrostemofoline (6) has been accomplished by preparing a key intermediate in the Overman synthesis of these alkaloids from commercially available 2-deoxy-D-ribose. The work presented in this account chronicles the evolution of our explorations to identify the optimal steric and electronic control elements necessary to generate the tricyclic core structure of these alkaloids in a single operation from an acyclic precursor. The key step in the synthesis is a novel cascade reaction sequence that is initiated by cyclization of a rhodium-derived carbene onto the nitrogen atom of a proximal imine group to generate an azomethine ylide that then undergoes spontaneous cyclization via dipolar cycloaddition. The synthesis features several other interesting reactions, including a Boord elimination to prepare a chiral allylic alcohol, a highly diastereoselective Hirama-Itô cyclization, and a useful modification of the Barton decarboxylation protocol. Further applications of cascade reactions to natural product synthesis are underway in our group, and will be reported in due course.

4. Experimental

4.1. General

Solvents and reagents were reagent grade and used without purification unless otherwise noted. Zn granules were activated by stirring with 1 M HCl for 10 min, filtering, rinsing with deionized H₂O, MeOH, then Et₂O, and drying under vacuum before use. MgO was dried by heating under vacuum at 140 °C overnight before use. Dichloromethane (CH₂Cl₂) and triethylamine (Et₃N) were distilled from calcium hydride and stored under nitrogen and methyl acetate was purified before each use by first drying over MgSO₄ and then distilling from P₂O₅. Tetrahydrofuran (THF) and diethyl ether (Et₂O) were passed through a column of neutral alumina and stored under argon. Methanol (MeOH) and dimethylformamide (DMF) were passed through a column of molecular sieves and stored under argon. Toluene was passed through a column of Q5 reactant and stored under argon. ¹H nuclear magnetic resonance (NMR) spectra were obtained at 400 or 500 MHz as indicated. Chemical shifts are reported in parts per million (ppm, δ) and referenced to the solvent. Coupling constants are reported in hertz (Hz). Spectral splitting patterns are designated as: s, singlet; d, doublet; t, triplet; m, multiplet; comp, overlapping multiplets of magnetically non-equivalent protons; br, broad; and bs, broad singlet. Infrared (IR) spectra were obtained using a Perkin-Elmer FTIR 1600 spectrophotometer on sodium chloride plates and reported as wavenumbers (cm⁻¹). Low-resolution chemical ionization mass spectra were obtained on a Finnigan TSQ-70 instrument, and high-resolution measurements were obtained on a VG Analytical ZAB2-E instrument. Analytical thin layer chromatography was performed using Merck 250 micron 60F-254 silica plates. The plates were visualized with UV light, p-anisaldehyde, and potassium permanganate. Flash column chromatography was performed according to Still’s method using ICN Silitech 32–63 D 60A silica gel.⁴³

4.2. (E)-Methyl 4-((4S,5S)-5-(iodomethyl)-2,2-dimethyl-1,3-dioxolan-4-yl)but-2-enoate (52)

Iodine (0.764 g, 3.01 mmol) was added to a solution of triphenylphosphine (0.781 g, 2.98 mmol) and imidazole (0.270 g, 3.96 mmol) in THF (15 mL) and the reaction was stirred at room temperature for 10 min. A solution of 51 (0.457 g, 1.98 mmol) in THF (6.5 mL) was added and the solution was heated under reflux for 3.5 h until complete consumption of 51 was observed by TLC. The reaction was cooled to room temperature, and poured into a stirred solution of aq. Na₂S₂O₃ (5% w/w, 25 mL). EtOAc (25 mL) was added, and the organic layer was removed. The aqueous layer was then extracted with CH₂Cl₂ (2 × 15 mL), and the combined organic layers were dried (Na₂SO₄), filtered, and concentrated. The crude residue was dissolved in Et₂O (50 mL) and the precipitate was removed by filtration and rinsed with an additional portion of Et₂O (50 mL). After concentrating the filtrate and washings, the crude residue was purified by flash chromatography eluting with hexanes:Et₂O (4:1) to give 0.490 g (73% yield) of 52 as a light yellow oil: ¹H NMR (CDCl₃, 400 MHz) δ 7.03-6.95 (m, 1 H), 5.95 (dt, J = 15.6, 1.2 Hz, 1 H), 4.41 (app q, J = 5.2 Hz, 1 H), 4.27 (app p, J = 4.0 Hz, 1 H), 3.73 (s, 3 H), 3.23-3.10 (m, 2 H), 2.56-2.42 (m, 2 H), 1.47 (s, 3 H), 1.36 (s, 3 H); ¹³C NMR (CDCl₃, 100 MHz) δ 166.2, 144.2, 123.2, 108.8, 77.8, 76.0, 51.4, 32.2, 28.2, 25.5, 2.3; IR (neat) 2987, 2948, 1721, 1660, 1436, 1381, 1041 cm⁻¹; HRMS (ESI) 363.0080 [C₁₁H₁₇INaO₄ (M+Na)⁺ requires 363.0064].

4.3. Methyl (2E,5S,6S)-5,6-bis(acetyloxy)-7-iodohept-2-enoate (54)

A solution of 2-deoxy-D-ribose 47 (25.0 g, 186.4 mmol) and methyl (triphenylphosphoranylidene) acetate 50 (74.8 g, 223.6 mmol) in THF (750 mL) was heated under reflux for 6 h and then cooled to room temperature. Imidazole (25.4 g, 372.7 mmol), PPh₃ (53.8 g, 205 mmol), and I₂ (54.4 g, 214.3 mmol) were then added sequentially to the reaction, and the mixture was stirred overnight in the dark at room temperature. The reaction was quenched with 10% aq. Na₂S₂O₃ (500 mL) and diluted with EtOAc (250 mL). The resulting layers were separated and the aq. layer extracted with CH₂Cl₂ (2 × 150 mL). The combined organic layers were dried (MgSO₄), filtered, and concentrated. The crude residue was dissolved in CH₂Cl₂ (375 mL), and then acetic anhydride (57.1 g, 559.1 mmol, 52.9 mL), DMAP (2.28 g, 18.7 mmol), and pyridine (44.2 g, 559.1 mmol, 45.2 mL) were added and the reaction was stirred at room temperature for 3 h. The reaction was then poured into a separatory funnel, and washed with 1 M aq. HCl (2 × 35 mL), saturated aqueous NaHCO₃ (1 × 35 mL), dried (MgSO₄), filtered and concentrated under reduced pressure. Trituration of the residue with Hex:Et₂O (100 mL, 8:1) precipitated out the Ph₃P=O byproduct formed during the olefination step. The precipitate was removed by filtration, and rinsed with Hex:Et₂O (150 mL, 8:1). The filtrate was then concentrated, redissolved in Et₂O (100 mL), and vacuum filtered through a 2.5 inch pad of SiO₂ using a 3 inch diameter fritted funnel. Once the initial filtrate had adsorbed onto the silica, the silica pad was rinsed with Et₂O (3 × 100 mL). The combined filtrate and washings were concentrated to give 62.3 g (87%) of 54 as a light yellow oil: ¹H NMR (CDCl₃, 400 MHz) δ 6.89-6.81 (m, 1 H), 5.88 (dt, J = 15.6, 1.2 Hz, 1 H), 5.18-5.14 (m, 1 H), 5.00-4.96 (m, 1 H), 3.73 (s, 3 H), 3.38-3.23 (m, 2 H), 2.61-2.47 (m, 2 H), 2.12 (s, 3 H), 2.07 (s, 3 H); ¹³C NMR (CDCl₃, 100 MHz) δ 169.50, 169.45, 165.9, 142.4, 123.9, 72.1, 71.5, 51.4, 32.4, 20.6, 20.6, 2.1; IR (neat) 2952, 1747, 1660, 1436, 1372, 1223, 1040 cm⁻¹; HRMS (ESI) m/z 406.9962 [C₁₂H₁₇INaO₆ (M+Na)⁺ requires 406.9962].

4.4. Methyl (2E,5S)-5-hydroxyhepta-2,6-dienoate (46)

A mixture of 54 (62.3 g, 162.2 mmol) and freshly activated Zn granules (53.0 g, 810.9 mmol) in anhydrous MeOH (635 mL) was heated under reflux for 16 h. After cooling the reaction to room temperature, the suspension was filtered through a pad of Celite then rinsed with MeOH (190 mL). The combined filtrate and washings were concentrated and the crude residue was purified by flash chromatography eluting with Hex:EtOAc (2:1) to give 15.7 g (62%) of 46 as a colorless oil: ¹H NMR (CDCl₃, 400 MHz) δ 7.01-6.93 (m, 1 H), 5.94-5.84 (comp, 2 H), 5.27 (dt, J = 17.6, 1.2 Hz, 1 H), 5.15 (dt, J = 10.4, 1.2 Hz, 1 H), 4.27 (q, J = 6.0 Hz, 1 H), 3.72 (s, 3 H), 2.72 (br s, 1 H), 2.45 (td, J = 7.2, 1.6 Hz, 2 H); ¹³C NMR (CDCl₃, 100 MHz) δ 166.8, 145.0, 139.7, 123.3, 115.3, 71.3, 51.4, 39.7; IR (neat) 3428, 2951, 1722, 1660, 1436 cm⁻¹; HRMS (CI) m/z 157.0864 [C₈H₁₃O₃ (M+1) requires 157.0865].

4.5. Methyl (2E,5S)-5-(carbamoyloxy)hepta-2,6-dienoate (55)

Chlorosulfonyl isocyanate (15.7 g, 110.5 mmol) was added dropwise to solution of 46 (15.7 g, 100.5 mmol) in CH₂Cl₂ (1 L) at 0 °C. The reaction was warmed to room temperature and stirred for 1 h, at which time H₂O (300 mL) was added. The reaction flask was then equipped with a short-path distillation apparatus and heated to 60 °C (bath temp) until all of the CH₂Cl₂ had distilled. The remaining aqueous mixture was extracted with EtOAc (3 × 25 mL), and the combined organic layers were washed with saturated aq. NaHCO₃ (1 × 25 mL), brine (1 × 25 mL), dried (Na₂SO₄), filtered, and concentrated to give 20.3 g (99%) of 55 as a light yellow oil: ¹H NMR (CDCl₃, 400 MHz) δ 6.90 (dt, J 7.2, 1 H), 5.90 (dt, J = 15.6, 1.2 Hz, 1 H), 5.80 (ddd, J = 16.8, 10.8, 6.4 Hz, 1 H), 5.33-5.20 (comp, 3 H), 4.94 (br s, 2 H), 3.74 (s, 3 H), 2.62-2.49 (comp, 2 H); ¹³C NMR (CDCl₃, 100 MHz) δ 166.5, 156.1, 143.4, 135.4, 123.8, 117.2, 73.3, 51.5, 36.9; IR (neat) 3474, 3364, 3203, 2953, 2925, 1714, 1660, 1604, 1438, 1384, 1320, 1041 cm⁻¹; HRMS (CI) m/z 200.0925 [C₉H₁₄NO₄ (M+H)⁺ requires 200.0923].

4.6. Methyl 2-[(4R,6S)-6-ethenyl-2-oxo-1,3-oxazinan-4-yl]acetate (45)

Compound 55 (20.3 g, 101.5 mmol) was dissolved in dry CH₂Cl₂ (895 mL) and cooled to −10 °C (bath temperature) in an ice/brine bath. NaH (4.31 g, 180.1 mmol, 60% w/w dispersion in mineral oil) was added in one portion, and the mixture was stirred under an atmosphere of N₂ (g) at −10 °C for 1.5 h. The reaction was quenched by the slow addition of saturated aqueous NH₄Cl (650 mL), and the aqueous layer was extracted with CH₂Cl₂ (4 × 300 mL). The combined organic layers were dried (MgSO₄), filtered, and concentrated under reduced pressure. The crude residue was then purified by recrystallization from methyl tert-butylether to give 17.09 g (85%) of 45 as a crystalline mixture (dr = 8:1) of diastereomers: ¹H NMR (CDCl₃, 400 MHz, major diastereomer) δ 6.48 (br s, 1 H), 5.92-5.83 (m, 1 H), 5.41 (d, J = 17.2 Hz, 1 H), 5.27 (d, J = 10.8 Hz, 1 H), 4.75 (q, J = 5.6 Hz, 1 H), 3.98-3.91 (m, 1 H), 3.72 (s, 3 H), 2.57 (dd, J = 4.4, 3.2 Hz, 2 H), 1.60-1.51 (m, 2 H); ¹³C NMR (CDCl₃, 100 MHz) δ 170.8, 153.8, 134.7, 117.4, 76.6, 51.9, 47.0, 39.9, 33.1; IR (neat) 3428, 2951, 1722, 1660, 1436 cm⁻¹; HRMS (CI) m/z 200.0924 [C₉H₁₄NO₄ (M+H) requires 200.0923].

4.7. tert-Butyl (4R,6S)-6-ethenyl-4-(2-methoxy-2-oxoethyl)-2-oxo-1,3-oxazinane-3-carboxylate (58)

A solution of 45 (17.1 g, 85.5 mmol), Boc₂O (37.4 g, 170.9 mmol), NEt₃ (26.0 g, 256.9 mmol, 258.1 mL), and DMAP (1.03 g, 8.55 mmol) in CH₂Cl₂ (430 mL) was stirred at room temperature overnight. The reaction was quenched with 1 M aq HCl (340 mL), and the layers were separated. The aqueous layer was then extracted with CH₂Cl₂ (2 × 500 mL), and the combined organic layers were dried (Na₂SO₄), filtered, and concentrated under reduced pressure. The crude residue was purified by flash chromatography eluting with hexanes:EtOAc (3:1) to give 22.2 g (87%) of syn-58 as a colorless oil: Major Diastereomer (syn-58): ¹H NMR (CDCl₃, 400 MHz) δ 5.82 (ddd, J = 17.2, 10.8, 6.0 Hz, 1 H), 5.38 (d, J = 17.2 Hz, 1 H), 5.26 (d, J = 10.8 Hz, 1 H), 4.67-4.61 (m, 1 H), 4.48 (dddd, J = 8.8, 8.8, 8.8, 2.8 Hz, 1 H), 3.66 (s, 3 H), 2.92 (dd, J = 16.0, 2.8 Hz, 1 H), 2.57 (dd, J = 16.0, 8.8 Hz, 1 H), 2.54-2.48 (m, 1 H), 2.74 (ddd, J = 14.0, 11.6, 8.8 Hz, 1 H), 1.49 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz) δ 170.5, 151.9, 150.6, 133.4, 118.3, 83.7, 76.0, 51.8, 50.3, 40.0, 34.8, 27.9; IR (neat) 2982, 2955, 1792, 1759, 1736, 1438, 1393, 1370, 1307, 1160 cm⁻¹; HRMS (CI) 300.1453 [C₁₄H₂₂NO₆ (M+H) requires 300.1447].

Minor diastereomer (anti-58): ¹H NMR (CDCl₃, 400 MHz) δ 5.87 (ddd, J = 17.6, 10.4, 6.0 Hz, 1 H), 5.41 (d, J = 17.6 Hz, 1 H), 5.30 (d, J = 10.4 Hz, 1 H), 4.93-4.87 (m, 1 H), 4.67 (ddd,= 15.6, J = 14.4, 6.0, 3.6 Hz, 1 H), 3.72 (s, 3H), 2.86 (dd, J = 16.0, 3.2 Hz, 1 H), 2.65 (dd, J = 16.0, 10.0 Hz, 1 H), 2.19 (app dt, J = 14.4, 3.2 Hz, 1 H), 2.08 (ddd, J = 14.4, 10.0, 5.2 Hz, 1 H), 1.54 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz) δ 170.4, 151.6, 148.5, 134.4, 117.9, 84.2, 74.9, 52.0, 49.6, 37.7, 31.2, 27.8; IR (neat) 2982, 2955, 1792, 1759, 1736, 1438, 1393, 1370, 1307, 1160 cm⁻¹; HRMS (CI) 300.1453 [C₁₄H₂₂NO₆ (M+H)⁺ requires 300.1447].

4.8. tert-Butyl (4R,6S)-2-oxo-6-vinyltetrahydro-2H-pyran-4-ylcarbamate (60)

A solution of 58 (0.150 g, 0.500 mmol) and Cs₂CO₃ (20 mg, 0.100 mmol) in MeOH (20 mL) was stirred at room temperature for 24 h. The reaction was concentrated and a slurry of the crude residue and SiO² (0.150 g) was created and the suspension was concentrated to dryness. The substrate adsorbed on silica was dried under high vacuum for 1 h, and then purified by column chromatography eluting with hexanes:EtOAc (2:1) to give 0.103 g (85%) of 60 as a colorless oil: ¹H NMR (CDCl₃, 400 MHz) δ 5.93-5.85 (m, 1 H), 5.36 (d, J = 17.2 Hz, 1 H), 5.29 (d, J = 9.6 Hz, 1 H), 5.08-5.03 (m, 1 H), 4.99 (br s, 1 H), 4.09 (br s, 1 H), 2.87 (dd, J = 17.6 Hz, 6.8 Hz, 1 H), 2.52 (dd, J = 17.6, 6.8 Hz, 1 H), 2.08-1.97 (m, 2 H), 1.47 (s, 9 H); LRMS (CI) m/z 241 [C₁₂H₁₉NO₄ (M+1) requires 241].

4.9. (5R,7S)-Methyl 5-(tert-butoxycarbonylamino)-2-diazo-7-hydroxy-3-oxonon-8-enoate (62)

A solution of 58 (0.388 g, 1.30 mmol) and Cs₂CO₃ (0.052 g, 0.259 mmol) in MeOH (52 mL) was stirred at room temperature for 24 h, and then the reaction was concentrated to dryness. In a separate flask, methyl acetate (0.577 g, 7.80 mmol) was added dropwise to a freshly prepared solution of LDA (7.80 mmol) in THF (7.4 mL) at −78 °C and the solution was stirred for 1 h. The enolate solution was then cannulated into a flask containing a solution of the crude residue from the previous step that had been precooled to 0 °C. The resulting solution was stirred at 0 °C for 1 h then it was warmed to room temperature and stirred for an additional 1 h. The reaction was quenched with saturated aqueous NH₄Cl (40 mL) and diluted with EtOAc (50 mL). The layers were separated and the aqueous layer was extracted with CH₂Cl₂ (2 × 50 mL). The combined organic layers were dried (Na₂SO₄), filtered, and concentrated, and the crude residue was purified by flash chromatography eluting with Hex:EtOAc (4:1) to give ~0.144 g of 61 as crude oil contaminated with 10–20% methyl acetoacetate. The crude residue, p-acetamidobenzenesulfonyl azide (p-ABSA, 0.127 g, 0.546 mmol), and NEt₃ (0.254 g, 2.52 mmol) in MeCN (5 mL) were stirred at room temperature overnight. The reaction mixture was concentrated, triturated with Et₂O (20 mL), filtered, and concentrated. The crude residue was purified by column chromatography eluting with Hex:EtOAc (1:1) to give 0.045 g (29%) of 62 as a yellow oil: ¹H NMR (CDCl₃, 400 MHz) δ 5.93-5.85 (m, 1 H), 5.27 (dt, J = 17.2, 1.2 Hz, 1 H), 5.12 (dt, J = 10.4, 1.2 Hz, 1 H), 5.08 (bs, 1 H), 4.25-4.22 (m, 1 H), 4.21-4.13 (m, 1 H), 3.84 (s, 3 H), 3.14 (dd, J = 15.6, 4.0 Hz, 1 H), 3.02 (dd, J = 15.8, 6.0 Hz, 1 H), 2.61 (bs, 1 H), 1.86-171 (comp, 2 H), 1.42 (s, 9 H); LRMS (CI) m/z 342 [C₁₅H₂₄N₃O₆ (M+1) requires 342].

4.10. Methyl (3R,5S)-3-{[(tert-butoxy)carbonyl]amino}-5-[(tert-butyldiphenylsilyl)oxy]hept-6-enoate (64)

A solution of syn-58 (22.22 g, 74.2 mmol) and Cs₂CO₃ (2.86 g, 14.8 mmol) in MeOH (370 mL) was stirred at room temperature for 48 h. The reaction was then concentrated under reduced pressure to provide 20.3 g of crude 59 as a colorless oil that was used in the next step without further purification. A solution of crude 59 (20.3 g, 74.2 mmol) in DMF (50 mL) was added to a solution of TBDPS-Cl (30.6 g, 111.3 mmol), imidazole (6.57 g, 96.46 mmol), and DMAP (0.091 g, 0.742 mmol) in DMF (320 mL), and the reaction was stirred at room temperature overnight. The reaction was quenched with 1 M aq HCl (300 mL), and the layers were separated. The aqueous layer was extracted with Et₂O (3 × 150 mL), and the combined organic layers were dried (MgSO₄), filtered, and concentrated under reduced pressure. The crude residue was purified by flash chromatography eluting with hexanes:EtOAc (5:1) to give 30.4 g (80%, from 58) of 64 as a colorless oil: ¹H NMR (CDCl₃, 400 MHz, rotamers) δ 7.70-7.63 (comp, 4 H), 7.47-7.33 (comp, 6 H), 5.85 (ddd, J = 17.2, 10.2, 6.4 Hz, 1 H), 5.10-5.03 (comp, 2 H), 4.41 (app d, J = 8.8 Hz, 1 H), 4.17 (app q, J = 6.0 Hz, 1 H), 3.91-3.81 (m, 1 H), 3.60 (s, 3 H), 2.40 (app d, J = 4.8 Hz, 2 H), 1.62 (app t, J = 6.4 Hz, 2 H), 1.36 (s, 9 H), 1.07 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz, rotamers) δ 171.8, 154.8, 139.4, 135.90, 135.87, 134.1, 133.7, 129.8, 129.6, 127.6, 127.4, 115.5, 79.0, 72.2, 51.5, 44.2, 42.3, 39.6, 28.3, 27.0, 19.2; IR (neat) 3423, 2959, 2932, 2858, 1737, 1716, 1502, 1170, 1111 cm⁻¹; HRMS (ESI) 512.2830 [C₂₉H₄₂NO₅Si (M+H)⁺ requires 512.2754].

4.11. Methyl (5R,7S)-5-{[(tert-butoxy)carbonyl]amino}-7-[(tert-butyldiphenylsilyl)oxy]-3-oxonon-8-enoate (65)

A solution of freshly distilled methyl acetate (4.75 g, 64.1 mmol) in THF (128 mL) was added dropwise via syringe pump to a solution of NaHMDS (83.33 mmol, 1.8 M in hexane) in THF (167 mL) at −78 °C. After 30 min, a solution of 64 (3.28 g, 6.41 mmol) in THF (13 mL) was added dropwise to the reaction via syringe pump. During the syringe pump additions the metal needle used to transfer the substrate solutions was passed through a −78 °C bath to precool the solutions before introduction into the reaction flask. After 1 h at −78 °C, the reaction was warmed to −10 °C (ice/brine bath) and stirred for 6 h. The reaction was then quenched by addition of saturated aqueous NH₄Cl (300 mL) and warmed to room temperature. The reaction mixture was extracted with EtOAc (5 × 100 mL), and the combined organic layers were dried (Na₂SO₄), filtered, and concentrated under reduced pressure. The crude residue was purified by flash chromatography eluting with hexanes:EtOAc (using a gradient from 9:1 to 5:1) to give 3.55 g (75%) of 65 as a colorless oil along with 0.56 g (17%) of recovered starting material 64: ¹H NMR (CDCl₃, 400 MHz, rotamers) δ 7.69-7.62 (comp, 4 H), 7.47-7.33 (comp, 6 H), 5.83 (ddd, J = 17.1, 10.4, 6.4 Hz, 1 H), 5.10-5.03 (comp, 2 H), 4.37 (app d, J = 8.0 Hz, 1 H), 4.15 (app q, J = 4.0 Hz, 1 H), 3.92-3.83 (m, 1 H), 3.70 (s, 3 H), 3.37 (s, 2 H), 2.61 (app d, J = 2.8 Hz, 2 H), 1.68-1.61 (m, 2 H), 1.35 (s, 9 H), 1.07 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz, rotamers) δ 172.6, 167.6, 154.9, 139.4, 135.9, 135.8, 134.0, 133.6, 129.8, 129.6, 127.6, 127.4, 115.5, 79.0, 72.3, 52.2, 49.0, 47.9, 43.9, 42.1, 28.2, 26.9, 19.1; IR (neat) 3417, 2957, 2932, 2858, 1746, 1715, 1714, 1502, 1246, 1169, 1111 cm⁻¹; HRMS (ESI) 576.2750 [C₃₁H₄₃NNaO₆Si (M+Na)⁺ requires 576.2757].

4.12. Methyl (5R,7S)-5-{[(tert-butoxy)carbonyl]amino}-7-[(tert-butyldiphenylsilyl)oxy]-2-diazo-3-oxonon-8-enoate (63)

A solution of 65 (3.55 g, 1.97 mmol), p-ABSA (0.708 g, 2.95 mmol), and NEt₃ (0.598 g, 5.91 mmol) in MeCN (6.6 mL) was stirred at room temperature for 16 h. The reaction was then concentrated and the crude residue was triturated with Et₂O (25 mL). The precipitate was filtered and rinsed with Et₂O/CH₂Cl₂ (25 mL, 2:1), and the combined filtrate and washings were concentrated under reduced pressure. The crude residue was purified by column chromatography eluting with Hex:EtOAc (2:1) to give 1.14 g (92%) of 63 as a yellow oil: ¹H NMR (CDCl₃, 400 MHz, rotamers) δ 7.71-7.63 (comp, 4 H), 7.46-7.33 (comp, 6 H), 5.83 (ddd, J = 17.0, 10.2, 6.4 Hz, 1H), 5.10-5.03 (comp, 2 H), 4.31 (app d, J = 9.6 Hz, 1 H), 4.21-4.13 (m, 1 H), 3.99-3.92 (m, 1 H), 3.81 (s, 3 H), 2.96-2.83 (comp, 2 H), 1.71-1.56 (comp, 2 H), 1.34 (s, 9 H), 1.06 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz, rotamers) δ 190.6, 161.6, 154.9, 139.4, 135.9, 134.1, 133.7, 129.7, 129.6, 128.2, 127.5, 127.4, 115.4, 78.8, 76.2, 72.2, 52.1, 45.6, 44.5, 42.7, 28.2, 26.9, 19.1; IR (neat) 3417, 2959, 2932, 2856, 2136, 1716, 1655, 1500, 1313, 1171, 1112 cm⁻¹; HRMS (ESI) 580.2838 [C₃₁H₄₂N₃O₆Si (M+H)⁺ requires 580.2843].

4.13. Methyl (5R)-5-[(2S)-2-[(tert-butyldiphenylsilyl)oxy]but-3-en-1-yl]-7-oxo-hexahydropyrrolo[1,2-c][1,3]oxazole-7a-carboxylate (67)

A mixture of 63 (0.190 g, 0.328 mmol) and Rh₂(OAc)₄ (0.007 g, 0.016 mmol) in CH₂Cl₂ (6.6 mL) was stirred at room temperature for 16 h. The reaction mixture was concentrated and the crude residue purified by column chromatography eluting with Hex:EtOAc (5:1) to give 0.181 g (86%) of 66 as a mixture (1:1) of diastereomers and rotamers. The crude residue and dimethoxymethane (0.138 g, 1.82 mmol) in TFA:CH₂Cl₂ (1:10, 8.6 mL) was stirred at room temperature for 6 h. The reaction was poured into a separatory funnel, diluted with CH₂Cl₂ (30 mL) and washed with saturated aq. NaHCO₃ (2 × 15 mL). The combined organic layers were dried (Na₂SO₄), filtered, and concentrated and the crude residue was then purified by column chromatography eluting with Hex:EtOAc (using a gradient elution from 1:0 to 6:1) to give 0.017 g (33%) of 67 as a mixture (1:1) of diastereomers as a colorless oil: ¹H NMR (CDCl₃, 400 MHz) δ 7.68-7.60 (comp, 4 H), 7.47-7.33 (comp, 6 H), 5.86-5.78 (m, 1 H), 5.05 (dt, J = 17.2, 1.2 Hz, 1 H), 5.03 (dt, J = 10.4, 1.2 Hz, 1 H), 4.51 (d, J = 7.2 Hz, 1 H), 4.23 (dt, J = 6.4, 4.8 Hz, 1 H), 4.15 (d, J = 7.2 Hz, 1 H), 4.08 (d, J = 9.2 Hz, 1 H), 3.92 (d, J = 9.2 Hz, 1 H), 3.73 (s, 3 H), 3.08-3.01 (m, 1 H), 2.26 (d, J = 7.2 Hz, 1 H), 2.25 (d, J = 10.0 Hz, 1 H), 1.98 (ddd, J = 13.8, 6.4, 3.6 Hz, 1 H), 1.72 (ddd, J = 13.8, 9.4, 4.8 Hz, 1 H), 1.07 (s, 9 H); LCMS (ESI) m/z 494.87 [C₂₈H₃₆NO₅Si (M+H) requires 494.67].

4.14. Methyl 3-[(tert-butyldiphenylsilyl)oxy]-9-oxo-7-azatricyclo [5.3.0.0^4,8]decane-8-carboxylate (71)

A solution of 67 (19 mg, 0.038 mmol) in toluene (3.8 mL) in a sealed tube reaction vessel was heated to 160 °C for 4 h. The reaction was cooled to rt, concentrated, and the crude residue was then purified by column chromatography eluting with Hex:EtOAc (1:1) to give 5.6 mg (32%) of 71 as a colorless oil along with 11 mg (62%) of 72 as a colorless oil: Minor Regioisomer (71): ¹H NMR (CDCl₃, 400 MHz) δ 7.62-7.58 (comp, 4 H), 7.46-7.34 (comp, 6 H), 3.96-3.91 (m, 1 H), 3.75 (s, 3 H), 3.55-3.52 (m, 1 H), 3.14 (ddd, J = 13.4, 11.4, 4.4 Hz, 1 H), 3.02 (dt, J = 8.4, 6.0 Hz, 1 H), 2.97 (dd, J = 6.2, 3.6 Hz, 1 H), 2.66-2.60 (m, 1 H), 2.36 (ddd, J = 13.4, 8.8, 4.4 Hz, 1 H), 1.84 (d, J = 18.0 Hz, 1 H), 1.71 (ddd, J = 18.2, 11.0, 5.2 Hz, 1 H), 1.65-1.57 (comp, 1 H), 1.37-1.32 (comp, 1 H), 1.05 (s, 9 H); LRMS (CI) m/z 464 [C₂₇H₃₄NO₄Si (M+H) requires 464].

Major Regioisomer (72): ¹H NMR (CDCl₃, 400 MHz) δ 7.64-7.60 (comp, 4 H), 7.46-7.35 (comp, 6 H), 3.85-3.79 (comp, 3 H), 3.74 (s, 3 H), 2.81 (dd, J = 15.8, 6.8 Hz, 1 H), 2.63-2.54 (comp, 2 H), 2.28-2.24 (m, 1 H), 2.05 (d, J = 15.6 Hz, 1 H), 1.66-1.60 (m, 1 H), 1.51 (d, J = 14.4 Hz, 1 H), 1.08 (s, 9 H), 0.9 (ddd, J = 15.6, 9.0, 5.2 Hz, 1H); LRMS (CI) m/z 464 [C₂₇H₃₄NO₄Si (M+H) requires 464].

4.15. Methyl (5R,7S)-5-[(E)-[2-(benzyloxy)-2-oxoethylidene]amino]-7-[(tert-butyldiphenylsilyl)oxy]-2-diazo-3-oxonon-8-enoate (73)

Trifluroroacetic acid (0.315 g, 2.76 mmol) was added to a precooled solution of 63 (0.160 g, 0.276 mmol) in CH₂Cl₂ (1.5 mL) at 0 °C. The reaction was then warmed to room temperature and stirred for 2 h. The reaction was then concentrated to dryness and pumped down under high vacuum for 2 h to ensure the removal of all excess TFA. 4 Å molecular sieves (0.100 g) were added to a solution of the crude residue in CH₂Cl₂ (1.5 mL) and the mixture was cooled to 0 °C. NEt₃ (0.028 g, 0.276 mmol) was then added dropwise, and upon completion of the addition, the reaction was warmed to room temperature. A 1 M solution of benzyl glyoxylate (0.41 mL, 0.414 mmol) in toluene was then added, and the reaction was stirred for 12 h at room temperature. The reaction mixture was then passed through a short pad of oven dried basic alumina, rinsing with anhyd. CH₂Cl₂ (10 mL), and the combined filtrate and washings were concentrated under reduced pressure to give 0.173 g (99%) of 73 as a colorless oil: ¹H NMR (CDCl₃, 400 MHz) δ 7.66 (s, 1 H), 7.63-7.59 (comp, 4 H), 7.39-7.28 (comp, 11 H), 5.75-5.67 (m, 1 H), 5.27 (s, 2 H), 4.93 (app d, J = 10.4 Hz, 1 H), 4.84 (app d, J = 17.2 Hz, 1 H), 4.10-4.05 (m, 1 H), 4.04-3.96 (m, 1 H), 3.79 (s, 3 H), 3.31 (dd, J = 18.0, 9.2 Hz, 1 H), 2.82 (dd, J = 16.0, 3.6 Hz, 1 H), 1.95-1.88 (m, 1 H), 1.84-1.77 (m, 1 H), 1.04 (s, 9 H).

4.16. 6-Benzyl-8-methyl-(6R)-3-[(tert-butyldiphenylsilyl)oxy]-9-oxo-7-azatricyclo [5.3.0.0^4,8]decane-6,8-dicarboxylate (76)

Trifluroroacetic acid (1.58 mL, 20.7 mmol) was added to a precooled solution of 63 (1.20 g, 2.07 mmol) in CH₂Cl₂ (10 mL) at 0 °C. The reaction was warmed to room temperature and stirred for 2 h. The reaction was then concentrated to dryness and pumped down under high vacuum for 2 h to ensure the removal of all excess TFA. 4 Å molecular sieves (1.2 g) were added to a solution of the crude residue in CH₂Cl₂ (6 mL) and the mixture was cooled to −20 °C. NEt₃ (0.32 mL, 2.28 mmol) was then added dropwise, and upon complete of the addition, the reaction was warmed to room temperature. A 1 M solution of benzyl glyoxylate (3.11 mL, 3.11 mmol) in CH₂Cl₂ was then added, and the reaction was stirred for 16 h at room temperature. The reaction was filtered through Celite and rinsed with CH₂Cl₂ (12 mL) to give 73 as a colorless oil along with an equimolar amount of NEt₃•TFA. The crude residue was dissolved in xylenes (40 mL), Rh₂(OAc)₄ (0.027 g, 0.062 mmol) was added, and the mixture was heated under reflux for 24 h. The reaction was concentrated to under reduced pressure, and the crude residue was purified by column chromatography eluting with Hex:EtOAc (3:1 to 1:1, with 1% v/v NEt₃) to give 0.93 g (75%) of 76 as a light yellow oil: ¹H NMR (CDCl₃, 400 MHz) δ 7.59-7.55 (comp, 4 H), 7.45-7.35 (comp, 11 H), 5.28 (d, J = 3.2 Hz, 2 H), 4.15 (app q, J = 6.0 Hz, 1 H), 4.08-4.04 (m, 1 H), 3.94-3.89 (m, 1 H), 3.72 (s, 3 H), 2.96 (app q, J = 3.2 Hz, 1 H), 2.68 (dd, J = 13.6, 5.6 Hz, 1 H), 2.61 (dd, J = 16.0, 6.4 Hz, 1 H), 2.15-2.07 (m, 1 H), 1.80 (d, J = 17.6 Hz, 1 H), 1.35-1.19 (comp, 2 H), 1.04 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz) δ 207.0, 172.5, 169.7, 167.4, 135.5, 135.5, 135.3, 134.7, 133.5, 133.0, 130.0, 129.9, 128.6, 128.5, 127.8, 127.6, 82.4, 67.0, 66.2, 62.3, 54.2, 53.1, 49.7, 44.0, 33.6, 27.0, 26.8, 19.0; IR (neat) 2953, 2857, 1738, 1741, 1428, 1228, 1112 cm⁻¹; HRMS (ESI) m/z 598.2614 [C₃₅H₄₀NO₆Si (M+H) requires 598.2625].

4.17. 6-Benzyl-8-methyl-(6R)-3-[(tert-butyldiphenylsilyl)oxy]-9-hydroxy-7-azatricyclo [5.3.0.0^4,8]decane-6,8-dicarboxylate (85)

A solution of 76 (0.085 g, 0.142 mmol) in MeOH (4.7 mL) was cooled to −30 °C and NaBH₄ (0.016 g, 0.426 mmol) was added in one portion. The reaction was stirred at −30 °C for 2 h then warmed slowly to room temperature over the course of 1 h. The reaction was quenched with aq. 1 M HCl (4 mL), and the resulting mixture was concentrated to remove all MeOH. The crude aq. mixture was neutralized with solid K₂CO₃, and then extracted with EtOAc (4 × 10 mL). The combined organic layers were dried (Na₂SO₄), filtered, and concentrated. The crude residue was purified by column chromatography eluting with 4:1 Hex:EtOAc to give 0.063 g (86%) of 85 as a colorless oil: ¹H NMR (CDCl₃, 400 MHz) δ 7.67-7.61 (comp, 4H), 7.44-7.31 (comp, 11 H), 5.24 (q, J = 13.2 Hz, 2 H), 4.77-4.72 (m, 1 H), 4.37 (dd, J = 10.4, 3.2 Hz, 1 H), 4.30 (dd, J = 11.2, 5.6 Hz, 1 H), 3.70 (s, 3 H), 3.67-3.62 (m, 1 H), 2.66 (dd, J = 14.0, 5.6 Hz, 1 H), 2.51 (dd, J = 6.0, 3.2 Hz, 1 H), 2.43-2.35 (m, 1 H), 1.97-1.88 (comp, 2 H), 1.64-1.57 (m, 1 H), 1.43-1.35 (m, 1 H), 1.33-1.09 (comp, 1 H), 1.06 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz) δ 173.3, 170.9, 135.75, 135.66, 135.6, 134.4, 134.2, 129.5, 129.5, 128.52, 128.48, 128.3, 127.4, 127.4, 80.5, 73.2, 66.7, 65.8, 61.6, 57.0, 52.4, 46.4, 37.4, 35.0, 27.3, 26.9, 19.1; IR (neat) 3233, 2955, 2892, 2857, 1736, 1471, 1225, 1108 cm⁻¹; HRMS (ESI) m/z 600.2777 [C₃₅H₄₂NO₆Si (M+H) requires 600.2781].

4.18. 4-Benzyl-2-methyl-(4R)-7-[(tert-butyldiphenylsilyl)oxy]-11-oxa-3-azatetracyclo [5.3.1.0^2,6.0^3,9]undecane-2,4-dicarboxylate (86)

A solution of 85 (32 mg, 0.0534 mmol), PhI(OAc)₂ (0.026 g, 0.0800 mmol), and I₂ (13.6 mg, 0.0534 mmol) in CH₂Cl₂ (2.7 mL) was irradiated with tungsten filament light bulb (150 W) at room temperature for 1.5 h. The reaction was quenched with 10% aq. sodium thiosulfate (3 mL), and extracted with EtOAc (4 × 5 mL). The combined organic layers were dried (MgSO₄), filtered, and concentrated. The crude residue was purified by column chromatography eluting with Hex:EtOAc (1:1) to give 29 mg (94 %) of 86 as a light yellow oil: ¹H NMR (CDCl₃, 400 MHz) δ 7.74-7.69 (comp, 4 H), 7.44-7.32 (comp, 11 H), 5.16 (app q, J = 8.4 Hz, 2 H), 4.64 (app t, J = 2.4 Hz, 1 H), 4.29 (app t, J = 9.2 Hz, 1 H), 3.69 (s, 3 H), 3.68-3.64 (m, 1 H), 2.73 (m, 1 H), 2.15-2.10 (comp, 2 H), 1.73-1.66 (comp, 1 H), 1.59 (d, J = 11.6 Hz, 1 H), 1.60-1.54 (comp, 1 H), 1.32 (d, J = 13.6 Hz, 1 H), 1.27-1.24 (m, 1 H), 1.05 (s, 9 H); ¹³C NMR (CDCl₃, 100 MHz) δ 171.2, 170.3, 136.0, 135.9, 135.4, 134.3, 134.2, 129.6, 129.6, 128.5, 128.4, 127.33, 127.29, 106.3, 85.0, 79.5, 77.2, 66.7, 63.7, 57.2, 56.9, 52.4, 37.9, 35.7, 29.4, 26.9, 19.1; IR (neat) 2955, 2858, 1736, 1457, 1214 cm^−1; HRMS (ESI) m/z 598.2623 [C₃₅H₄₀NO₆Si (M+H) requires 598.2625].

4.19. Methyl 7-[(tert-butyldiphenylsilyl)oxy]-11-oxa-3-azatetracyclo [5.3.1.0^2,6.0^3,9]undecane-2-carboxylate (89)

A suspension of benzyl ester 86 (0.419 g, 0.701 mmol) and 10 % w/w Pd/C (112 mg) in EtOH (14 mL) was stirred under an atmosphere of H₂ (gas) at room temperature for 16 h. The mixture was filtered through a short pad of Celite, which was rinsed with EtOH (30 mL), and the combined filtrate and washings were concentrated to dryness to give 0.358 g (>99%) of acid 87 as an amorphous solid. A mixture of crude acid 87 (0.193 g, 0.424 mmol), 1-hydroxypyridine-2(1H)-thione (88, 0.073 g, 0.570 mmol), DCC (0.118 g, 0.570 mmol), and DMAP (0.047 g, 0.424 mmol) was dissolved in CHCl³ (4.0 mL). The resulting canary yellow solution was treated with t-BuSH (0.43 mL, 0.424 mmol) and immediately irradiated with a tungsten filament light bulb (250 W) at room temperature for 1 h. The reaction was then concentrated, and the crude residue was directly purified by column chromatography eluting with hexanes:EtOAc (as a gradient from 3:1 → 1:1 → 1:5 with 1% v/v Et₃N) to give 0.126 g (71%) of 89 as a yellow oil: ¹H NMR (CDCl₃, 300 MHz) δ 7.77-7.72 (comp, 4 H), 7.45-7.34 (comp, 6 H), 4.65 (s, 1 H), 3.71 (s, 3 H), 3.23 (m, 1 H), 3.13 (ddd, J = 12.9, 9.0, 6.0 Hz, 1 H), 2.93 (ddd, J = 13.5, 7.8, 5.7 Hz, 1 H), 2.78-2.76 (m, 1 H), 1.83-1.71 (comp, 2 H), 1.65-1.60 (comp, 3 H), 1.39 (d, J = 16.2 Hz, 1 H), 1.06 (s, 9 H); ¹³C NMR (CDCl₃, 75 MHz) δ 172.0, 136.0, 136.0, 134.6, 134.4, 129.6, 129.6, 127.4, 127.3, 106.3, 83.9, 80.3, 61.1, 60.4, 58.3, 52.3, 49.5, 38.1, 35.4, 26.9, 26.6, 21.0, 19.1, 14.2; IR (neat) 2953, 1740, 1430, 1276, 1218, 1109 cm⁻¹; HRMS (ESI) m/z 464.2256 [C₂₇H₃₄NO₄Si (M+H) requires 464.2252].

4.20. 2-[(1E)-but-1-en-1-yl]-11-oxa-3-azatetracyclo [5.3.1.0^2,6.0^3,9]undecan-7-ol (95)

DIBAL-H (0.41 mL, 1.0 M in hexanes, 0.41 mmol) was added dropwise to solution of 89 (0.126 g, 0.272 mmol) in CH₂Cl₂ (5 mL) at −78 °C and the reaction was stirred for 3 h. The reaction was quenched with MeOH (0.3 mL) and half saturated potassium sodium tartrate solution (5 mL) at −78 °C, and the reaction mixture was warmed to room temperature and stirred vigorously until the organic layer became clear. The separated aqueous layer was extracted with CH₂Cl₂ (3 × 10 mL), and the combined organic layers were dried (Na₂SO₄), filtered, and then concentrated. The crude residue was purified by column chromatography eluting with EtOAc:MeOH (5:1) with 1% v/v Et₃N to give 0.096 g (81%) of 92 as a light yellow oil.

To a stirred solution of 92 (0.096 g, 0.221 mmol) and 1-phenyl-1H-tetrazol-5-yl sulfone 93 (0.168 g, 0.664 mmol) in DME (7.4 mL) at −55 °C was added KHMDS (1.8 mL, 0.5 M in toluene, 0.884 mmol) dropwise. The resulting solution was stirred for 1 h at −55 °C and warmed to room temperature. After stirring for 1 h at room temperature, the reaction mixture was quenched with saturated NaCl solution (5 mL). The separated aqueous layers were extracted with EtOAc (3 × 10 mL), and the combined organic layers were dried (Na₂SO₄), filtered, and then concentrated to give ~0.102 g of 94 as a crude oil. To a stirred solution of crude 94 (~0.102 g, 0.221 mmol) in THF (4.4 mL) at room temperature was added TBAF (0.209 g, 0.663 mmol). The resulting solution was stirred for 4 h at room temperature before it was filtered through a pad of silica gel, and washed with CH₂Cl₂:MeOH (2:1, 40 mL) containing 1% v/v Et₃N. The combined solution was concentrated, and the crude residue was purified by column chromatography eluting with hexanes:EtOAc (1:1) to remove the nonpolar impurities followed by EtOAc:MeOH (5:1) with 1% v/v Et₃N to give 0.046 g (94%) of 95 as a light yellow oil: ¹H NMR (CDCl₃, 300 MHz) δ 5.70 (dt, J = 6.0 Hz, 1 H), 4.48 (d, J = 15.6 Hz, 1 H), 4.30 (d, J = 1.8 Hz, 1 H), 3.39 (m, 1 H), 3.30 (comp, 2 H), 2.38 (m, 1 H), 2.09-2.04 (comp, 2 H), 1.86-1.80 (comp, 5 H), 1.62 (d, J = 13.2 Hz, 1 H), 0.99 (t, J = 7.5 Hz, 3 H); ¹³C NMR (CDCl₃, 75 MHz) δ 132.4, 127.4, 105.0, 82.7, 82.4, 60.8, 56.7, 48.2, 37.8, 34.5, 26.5, 25.4, 13.6; IR (neat) 3337, 2959, 2878, 1350, 1328, 1056, 977 cm⁻¹; HRMS (ESI) m/z 222.1487 [C₁₃H₂₀NO₂ (M+H) requires 222.1489].

4.21. 6-Benzyl-8-methyl-(6R)-3-[(tert-butyldiphenylsilyl)oxy]-9-methyloxymethyloxy-7-azatricyclo[5.3.0.0^4,8]decane-6,8-dicarboxylate (97)

To a stirred solution of 85 (0.624 g, 2.04 mmol) in DMF (10 mL) was sequentially added MOM-Cl (0.79 mL, 10.4 mmol) and N(i-Pr)₂Et (3.62 mL, 20.8 mmol) at room temperature. The resulting solution was heated to 50 °C and stirred for 16 h at 50 °C. The reaction was quenched with MeOH (1.0 mL), diluted with H₂O (50 mL), and extracted with EtOAc (4 × 20 mL). The combined organic layers were dried (Na₂SO₄), filtered, concentrated, and the crude residue was purified by column chromatography eluting with hexanes:EtOAc (as a gradient from 2:1 to 1:2) with 1% v/v Et₃N to give 0.502 g (75%) of 97 as a light yellow oil: ¹H NMR (CDCl₃, 300 MHz) δ 7.69-7.61 (comp, 4 H), 7.41-7.32 (comp, 11 H), 5.24 (q, J = 12.0 Hz, 2 H), 4.68-4.62 (m, 1 H), 4.28 (d, J = 1.2 Hz, 2 H), 4.32-4.26 (m, 1 H), 4.03 (dd, J = 11.1, 6.0 Hz, 1 H), 3.72-3.71 (m, 1 H), 3.71 (s, 3 H), 3.06 (s, 3 H), 2.79 (dd, J = 6.3, 3.3 Hz, 1 H), 2.61 (dd, J = 13.8, 5.7 Hz, 1 H), 2.48-2.38 (m, 1 H), 2.08-1.99 (m, 1 H), 1.33-1.09 (comp, 3 H), 1.05 (s, 9 H); ¹³C NMR (CDCl₃, 75 MHz) δ 173.0, 170.8, 135.8, 135.7, 134.6, 134.1, 129.5, 128.5, 128.5, 128.3, 127.4, 96.1, 80.0, 78.3, 76.6, 66.7, 66.1, 61.3, 57.1, 55.3, 52.5, 45.9, 37.4, 34.9, 27.6, 26.9, 19.1; IR (neat) 2953, 2893, 2857, 1739, 1225, 1111, 1040, 702 cm⁻¹; HRMS (ESI) m/z 644.3037 [C₃₇H₄₆NO₇Si (M+H) requires 644.3038].

4.22. Methyl 3-[(tert-butyldiphenylsilyl)oxy]-9-methyloxymethyloxy-7-azatricyclo[5.3.0.0^4,8]-decane-8-carboxylate (98)

A suspension of 97 (0.340 g, 0.528 mmol) and 10 % w/w Pd/C (84 mg) in EtOH (11 mL) was stirred under an atmosphere of H₂ (gas) at room temperature for 16 h. The mixture was filtered through a short pad of Celite, which was rinsed with EtOH (30 mL), and the combined filtrate and washings were concentrated to dryness to give 0.292 g (100%) of the corresponding carboxylic acid as an amorphous solid. The acid thus obtained (0.292 g, 0.528 mmol), 1-hydroxypyridine-2(1H)-thione (88, 0.101 g, 0.881 mmol), DCC (0.163 g, 0.881 mmol), and DMAP (0.064 g, 0.528 mmol) were dissolved in CHCl₃ (5.3 mL). t-BuSH (0.59 mL, 5.28 mmol) was added to the solution, and the solution was immediately irradiated with a tungsten filament light bulb (250 W) at room temperature for 1 h. The reaction was concentrated, and the residue was purified by column chromatography eluting with hexanes:EtOAc (as a gradient from 2:1 to 0:1) with 1% v/v Et₃N to give 0.170 g (63%) of 98 as a yellow oil: ¹H NMR (CDCl₃, 300 MHz) δ 7.71-7.66 (comp, 4 H), 7.42-7.34 (comp, 6 H), 4.72-4.64 (m, 1 H), 4.36-4.29 (comp, 2 H), 3.72 (s, 3 H), 3.22-3.18 (m, 1 H), 3.10 (s, 3 H), 2.93 (dd, J = 10.5, 4.2 Hz, 1 H), 2.84 (dd, J = 8.4, 5.4 Hz, 1 H), 2.77 (dd, J = 5.7, 3.9 Hz, 1 H), 2.48-2.38 (m, 1 H), 2.31-2.22 (m, 1 H), 1.66-1.58 (m, 1 H), 1.52-1.48 (m, 1 H), 1.41-1.35 (comp, 2 H), 1.13 (dd, J = 16.2, 2.4 Hz, 1 H), 1.07 (s, 9 H); ¹³C NMR (CDCl₃, 75 MHz) δ 173.7, 135.9, 135.8, 134.8, 134.3, 129.5, 127.4, 127.4, 96.0, 79.3, 78.8, 77.2, 66.4, 60.3, 55.3, 52.4, 47.4, 46.6, 37.7, 34.7, 27.0, 24.7, 19.2; IR (neat) 2951, 2889, 2857, 1737, 1428, 1262, 1229, 1107, 1044 cm⁻¹; HRMS (ESI) m/z 510.2673 [C₂₉H₄₀NO₅Si (M+H) requires 510.2670].

4.23. 3-[(tert-butyldiphenylsilyl)oxy]-9-(methyloxymethyloxy)-7-azatricyclo[5.3.0.0^4,8]decane-8-carbaldehyde (99)

DIBAL-H (0.74 mL, 1.0 M in hexanes, 0.74 mmol) was added dropwise to solution of 98 (0.251 g, 0.492 mmol) in CH₂Cl₂ (10 mL) at −78 °C and the reaction was stirred for 2 h. The reaction was quenched with MeOH (0.5 mL) and half saturated potassium sodium tartrate solution (5 mL) at −78 °C, and the reaction mixture was warmed to room temperature and stirred vigorously until the organic layer became clear. The separated aqueous layer was extracted with CH₂Cl₂ (2 × 10 mL), and the combined organic layers were dried (Na₂SO₄), filtered, and then concentrated. The crude residue was purified by column chromatography eluting with hexanes:EtOAc (as a gradient from 1:1 to 1:5) with 1% v/v Et₃N to give 0.212 g (90%) of 99 as a light yellow oil: ¹H NMR (CDCl₃, 300 MHz) δ 9.30 (s, 1 H), 7.70-7.65 (comp, 4 H), 7.42-7.33 (comp, 6 H), 4.61 (m, 1 H), 4.30-4.23 (comp. 3 H), 3.21-3.19 (m, 1 H), 3.04 (s, 3 H), 2.88-2.73 (comp, 2 H), 2.62 (dd, J = 5.7, 3.3 Hz, 1 H), 2.48-2.38 (m, 1 H), 2.26 (ddd, J = 13.5, 9.0, 5.1 Hz, 1 H), 1.62-1.52 (comp, 2 H), 1.47-1.40 (m, 1 H), 1.32-1.25 (m, 1 H), 1.21 (d, J = 13.8, 2.7 Hz, 1 H), 1.07 (s, 9 H); ¹³C NMR (CDCl₃, 75 MHz) δ 201.5, 135.9, 135.8, 127.5, 127.4, 96.0, 82.3, 75.4, 66.3, 61.0, 55.3, 47.7, 43.2, 38.2, 35.3, 27.0, 24.7, 19.2; IR (neat) 2933, 2887, 2857, 1731, 1111, 1088, 1045, 703 cm⁻¹; HRMS (ESI) m/z 480.2562 [C₂₈H₃₈NO₄Si (M+H) requires 480.2565].

4.24. 8-[(1E)-but-1-en-1-yl]-3-[(tert-butyldiphenylsilyl)oxy]-9-(methyloxymethyloxy)-7-azatricyclo[5.3.0.0^4,8]decane (100)

KHMDS (3.52 mL, 0.5 M in toluene, 11.76 mmol) was added dropwise to a solution of 99 (0.212 g, 0.44 mmol) and 1-phenyl-5-propylsulfonyl-1H-tetrazole (0.335 g, 1.33 mmol) in DME (15 mL) at −55 °C and the reaction was stirred for 1 h at −55 °C then for 1 h at room temperature. The reaction mixture was quenched with saturated aq NaCl solution (5 mL), the layers separated, and the aqueous layer was extracted with EtOAc (3 × 10 mL). The combined organic layers were dried (Na₂SO₄), filtered, concentrated, and the crude residue was purified by column chromatography eluting with hexanes:EtOAc (as a gradient from 3:1 → 1:1 → 0:1) with 1% v/v Et₃N to give 0.240 g (89%) of 100 as colorless oil: ¹H NMR (CDCl₃, 300 MHz) δ 7.72-7.66 (comp, 4 H), 7.41-7.33 (comp, 6 H), 5.54 (dt, J = 15.6, 6.3 Hz, 1 H), 5.26 (d, J = 15.6 Hz, 1 H), 4.66-4.60 (m, 1 H), 4.35 (d, J = 6.6 Hz, 1 H), 4.32 (d, J = 6.6 Hz, 1 H), 3.95 (dd, J = 10.5, 3.0 Hz, 1 H), 3.14-3.07 (m, 1 H), 3.09 (s, 3 H), 2.88-2.68 (m, 3 H), 2.36-2.29 (m, 1 H), 2.22 (dd, J = 6.0, 3.6 Hz, 1 H), 2.12 (ddd, J = 12.9, 8.4, 4.5 Hz, 1 H), 2.00 (dt, J = 7.5, 1.5 Hz, 2 H), 1.61-1.51 (m, 2 H), 1.42-1.36 (m, 1 H), 1.07 (s, 9 H), 0.95 (t, J = 7.2 Hz, 3 H),; ¹³C NMR (CDCl₃, 75 MHz) δ 135.9, 135.8, 132.4, 130.9, 129.3, 127.4, 96.0, 82.1, 75.7, 66.9, 60.3, 55.1, 47.4, 47.0, 37.3, 35.2, 27.0, 25.5, 24.8, 19.2, 13.7; IR (neat) 2958, 2932, 2886, 2857, 1472, 1428, 1106, 1043, 703 cm⁻¹; HRMS (ESI) m/z 506.3092 [C₃₁H₄₄NO₃Si (M+H) requires 506.3085].

4.25. 8-[(1E)-but-1-en-1-yl]-9-(methyloxymethyloxy)-7-azatricyclo[5.3.0.0^4,8]decan-3-ol (101)

A solution of 100 (0.130 g, 0.256 mmol), and TBAF (0.487 g, 1.54 mmol) in THF (5.0 mL) was stirred at 50 °C for 16 h. The reaction was cooled to room temperature and Et₃N (0.2 mL) was added. The reaction was concentrated and the crude residue was purified by column chromatography eluting with hexanes:EtOAc (3:1) to remove the nonpolar impurities and then with EtOAc/MeOH (10:1) with 1% v/v Et₃N to give 0.065 g (95%) of 101 as a light yellow oil: ¹H NMR (CDCl₃, 300 MHz) δ 5.68 (dt, J = 15.6, 6.6 Hz, 1 H), J = 15.6 5.48 (d, Hz, 1 H), 4.65-4.54 (comp, 3 H), 4.14 (dd, J = 10.5, 2.7 Hz, 1 H), 3.35 (s, 3 H), 3.29-3.26 (m, 1 H), 2.91-2.82 (m, 1 H), 2.71 (ddd, J = 14.1, 8.7, 5.4 Hz, 1 H), 2.55-2.45 (m, 1 H), 2.32 (dd, J = 6.0, 3.6 Hz, 1 H), 2.09-1.97 (m, 2 H), 1.85 (ddd, J = 13.2, 8.7, 5.4 Hz, 1 H), 1.66-1.56 (m, 2 H), 1.45-1.36 (m, 2 H), 0.97 (t, J = 7.5 Hz, 3 H); ¹³C NMR (CDCl₃, 75 MHz) δ 132.1, 131.5, 96.4, 82.1, 75.8, 65.0, 60.2, 55.4, 47.3, 46.9, 37.3, 34.3, 25.5, 24.6, 13.7; IR (neat) 3368, 2957, 2886, 1454, 1151, 1097, 1042, 961 cm⁻¹; HRMS (ESI) m/z 268.1910 [C₁₅H₂₆NO₃ (M+H) requires 268.1907].

4.26. 8-[(1E)-but-1-en-1-yl]-9-(methyloxymethyloxy)-7-azatricyclo[5.3.0.0^4,8]decan-3-one (105)

Et₃N (0.068 mL, 0.486 mmol) and SO₃·Py (0.039 g, 0.243 mmol) were sequentially added to a solution of 101 (0.013 g, 0.049 mmol) in CH₂Cl₂ (1 mL) and DMSO (0.5 mL) at room temperature and the reaction was stirred for 4 h. The reaction was diluted with saturated aq. NaHCO₃ solution (2 mL), the layers separated, and the aqueous layer was extracted with EtOAc (3 × 5 mL). The combined organic layers were dried (Na₂SO₄), filtered, and concentrated, and the crude residue was purified by column chromatography eluting with hexanes:EtOAc (as a gradient from 1:1 to 0:1) with 1% v/v Et₃N to give 0.010 g (77%) of 105 as colorless oil: ¹H NMR (CDCl₃, 300 MHz) δ 5.74 (dt, J = 15.9, 6.9 Hz, 1 H), J = 15.9, 1.5 Hz, 1 H), 4.59 (s, 2 H), 4.21 (d, J = 8.7 Hz, 1 H), 3.53 (t, J = 6.3 Hz, 1 H), 3.33 (s, 3 H), 3.16-3.00 (m, 2 H), 2.85 (d, J = 6.6 Hz, 1 H), 2.58-2.50 (m, 1 H), 2.45-2.36 (m, 1 H), 2.19-2.11 (m, 1 H), 2.07 (comp, 3 H), 1.69-1.60 (comp, 2 H), 0.99 (t, J = 7.5 Hz, 3 H); ¹³C NMR (CDCl₃, 75 MHz) δ 209.3, 132.2, 129.9, 95.8, 82.0, 79.7, 60.4, 56.9, 55.6, 46.6, 41.2, 38.6, 30.9, 25.5, 13.6; IR (neat) 2958, 2893, 1722, 1106, 1091, 1040 cm⁻¹; HRMS (ESI) m/z 266.1754 [C₁₅H₂₄NO³ (M+H) requires 266.1751].

4.27. Ethyl 2-{8-[(1E)-but-1-en-1-yl]-9-(methyloxymethyloxy)3-oxo-7-azatricyclo[5.3.0.0^4,8]decan-2-yl}acetate (106)

Freshly prepared LDA (0.42 mL, 0.2 M in THF, 0.084 mmol) was added to a solution of 105 (0.016 g, 0.060 mmol) in THF (1.2 mL) at −10 °C, and the reaction was stirred for 1 h at −10 °C. Ethyl iodoacetate (0.011 mL, 0.090 mmol) was then added at −10 °C, and the reaction was continued for 30 min. DABCO (20 mg) was added, and the solution was warmed to room temperature. Saturated aq. NaCl (5 mL) and EtOAc (5 mL) were added, the layers were separated, and the aqueous layer extracted with EtOAc (2 × 10 mL). The combined organic layers were dried (Na₂SO₄), filtered, and then concentrated under reduced pressure, and the crude residue was purified by column chromatography eluting with hexanes:EtOAc (as a gradient from 1:1 to 0:1) with 1% v/v Et₃N to give 0.013 g (62%) of 106 as colorless oil along with 0.003 g (16%) of starting material 105: ¹H NMR (CDCl₃, 300 MHz) δ 5.72 (dt, J = 15.6, 6.3 Hz, 1 H), J = 15.6 Hz,5.41 (d, 1 H), 4.57 (s, 2 H), 4.22-4.14 (comp, 3 H), 3.40 (d, J = 5.7 Hz, 1 H), 3.33 (s, 3 H), 3.18-3.07 (m, 2 H), 2.88-2.75 (comp. 3 H), 2.54 (dd, J = 15.6, 10.8 Hz, 1 H), 2.41-2.32 (m, 1 H), 2.19-2.02 (comp, 3 H), 1.74 (d, J = 12.9 Hz, 1 H), 1.70-1.62 (m, 1 H), 1.27 (t, J = 7.2 Hz, 3 H), 0.99 (t, J = 7.5 Hz, 3 H); ¹³C NMR (CDCl₃, 75 MHz) δ 209.4, 171.7, 132.3, 129.3, 95.6, 82.2, 78.9, 63.0, 60.9, 56.6, 55.7, 48.3, 45.7, 41.8, 38.9, 30.2, 25.5, 14.2, 13.6; IR (neat) 2960, 1732, 1715, 1151, 1106, 1039 cm^−1; HRMS (ESI) m/z 352.2123 [C₁₉H₃₀NO₅ (M+H) requires 352.2119].

4.28. Ethyl 2-{2-[(1E)-but-1-en-1-yl]-7-hydroxy-11-oxa-3-azatetracyclo[5.3.1.0^2,6.0^3,9]undecan-8-yl}acetate (107)

A solution of 106 (0.014 g, 0.040 mmol) and DBU (0.024 mL, 0.16 mmol) in toluene (0.4 mL) was heated at 130 °C (bath temperature) for 4 h in a screw capped vial. The reaction was cooled to room temperature and filtered through a pad of silica gel (first with EtOAc then EtOAc:MeOH 10:1) to afford the 0.014 g of the epimerized product as yellow oil. The crude residue was dissolved in CH₂Cl² (0.4 mL) and TFA (0.31 mL, 4.0 mmol) was added dropwise at 0 °C. The reaction was warmed to room temperature and stirred for 1 h. The reaction was sequentially diluted with 5 M aq. NaOH (1 mL), CH₂Cl² (5 mL), and NaHCO³ (5 mL) and the layers were separated. The aqueous layer was extracted with CH₂Cl₂ (3 × 5 mL), and the combined organic layers were dried (Na₂SO₄), filtered, and concentrated. The crude residue was purified by column chromatography eluting with EtOAc with 1% v/v Et₃N to give 0.010 g (81%) of 107 as light yellow oil: ¹H NMR (CDCl₃, 500 MHz) δ 5.72 (dt, J = 15.5, 6.0 Hz, 1 H), J = 15.55.48 (d, Hz, 1 H), 5.36 (br s, 1 H), 4.25 (br s, 1 H), 4.19 (qd, J = 7.0, 5.41 (dt,2.0 Hz, 2 H), 3.19 (br s, 1 H), 3.06-2.96 (m, 2 H), 2.92 (dd, J = 17.0, 9.5 Hz, 1 H), 2.45 (t, J = 3.5 Hz, 1 H), 2.23-2.19 (m, 2 H), 2.09-2.03 (m, 2 H), 1.89-1.81 (comp, 3 H), 1.64 (dt, J = 12.0, 3.0 Hz, 1 H), 1.28 (t, J = 7.0 Hz, 3 H), 0.99 (t, J = 7.5 Hz, 3 H); ¹³C NMR (CDCl₃, 125 MHz) δ 175.6, 132.6, 127.2, 104.8, 82.3, 81.6, 65.3, 61.5, 56.0, 47.8, 36.9, 33.2, 32.5, 26.8, 25.4, 14.1, 13.6; IR (neat) 3349, 2962, 1733, 1325, 1273, 1227, 1179, 1038, 972 cm⁻¹; HRMS (ESI) m/z 308.1859 [C₁₇H₂₆NO⁴ (M+H) requires 308.1856]; [α]_D²⁵ +17.3 (c = 0.5, CHCl₃).

Scheme 19.

Advancing Cycloadduct to a Formal Total Synthesis