Two approaches to the synthesis of compounds corresponding to the C17–C27 fragment of the 20-deoxybryostatins are described. The first approach is based on the palladium(0) catalysed coupling of tin enolates, generated in situ from enol acetates using tributyltin methoxide, with vinylic bromides. The vinylic bromides were prepared using the Sharpless asymmetric dihydroxylation to introduce the hydroxyl groups corresponding to those at C25 and C26 in the bryostatins. Following several steps to introduce alkynyl ester functionality, the stereoselective addition of a tributyltin cuprate followed by tributyltin–bromine exchange gave the required vinylic bromides. The palladium(0) catalysed couplings worked very well for enol esters containing thioether substituents and gave products with retention of the position and geometry of the trisubstituted double bond derived from the vinylic bromide. These were taken through to compounds corresponding to fully developed C17–C27 fragments ready for assembly of the 16,17-double-bond of bryostatins by Julia reactions. This chemistry was also applied to prepare intermediates suitable for incorporation into bryostatins by ring-closing metathesis but, in this case, the coupling reaction gave mixtures of products including both the required βγ-unsaturated ketone and a conjugated diene formed by a competing Heck reaction. To avoid this problem, a second approach to compounds suitable for incorporation into a metathesis-based assembly of 20-deoxybryostatins was developed. In this organotin-free synthesis, the key step was the conjugate addition of an organic cuprate generated from allylmagnesium bromide to an alkynoate that gave the required (Z)-trisubstituted alkene with excellent stereoselectivity. This was converted into metathesis precursors in a few steps.

The first total synthesis of a derivative of a 20-deoxybryostatin, namely 7-des-O-pivaloyl-7-O-benzylbryostatin 10 is described. Preliminary studies demonstrated that the modified Julia reactions of 2-benzothiazolylsulfones corresponding to the C17–C27 fragment with aldehydes corresponding to the C1–C16 fragment, provided an efficient and stereoselective assembly of advanced intermediates with the (E)-16,17-double-bond. The synthesis of the C1–C16 fragment was then modified so that the C1 acid was present as its allyl ester before the Julia coupling. A more efficient synthesis of the C17–C27 sulfone was developed in which a key step was the bismuth mediated coupling of an allylic bromide with an aldehyde in the presence of an acrylate moiety in the allylic bromide. A scalable synthesis of an advanced macrolide was completed using the modified Julia reaction followed by selective deprotection and macrolactonisation. The final stages of the synthesis required selective hydroxyl deprotection and the introduction of the sensitive C19–C21 unsaturated keto-ester functionality. Unexpected selectivities were observed during studies of the hydroxyl group deprotections. In particular, cleavage of tri-isopropylsilyl ethers of the exocyclic primary allylic alcohols was observed in the presence of the triethylsilyl ether of the secondary alcohol at C19. Model studies helped in the design of the methods used to introduce the C19–C21 keto-ester functionality and led to the completion of a total synthesis of a close analogue of bryostatin 10 in which a benzyloxy group rather than the pivaloyloxy group was present at C7.

The modified Julia reaction and acyl carbanion chemistry, especially reactions of 2-lithiated dithianes, have been investigated for the synthesis of intermediates that are the synthetic equivalents of the C11–C27 fragments of bryostatins. The modified Julia reaction using 2-benzothiazolylsulfones was found to be more useful for the formation of the C16–C17 double-bond than the classical Julia reaction using phenylsulfones, and bulky sulfones gave very good (E)-stereoselectivity. The alkylation of a dithiane monoxide that corresponded to a C19-acyl carbanion using (E)-1-bromobut-2-ene was efficient but the use of a more complex allylic bromide corresponding to the C20–C27 fragment of the bryostatins was unsuccessful, possibly due to competing elimination reactions. This meant that the use of C19 dithianes for the synthesis of 20-deoxybryostatins would have to involve the stepwise assembly of the C20–C27 fragment from simpler precursors. However, lithiated C19 dithianes gave good yields of adducts with aldehydes and conditions were developed for the stereoselective conversion of the major adducts into methoxyacetals that corresponded to the C17–C27 fragment of 20-oxygenated bryostatins. A convergent synthesis of the C11–C27 fragment of a 20-deoxybryostatin was subsequently achieved using a 2-benzothiazolylsulfone corresponding to the intact C17–C27 fragment.

Vinylic iodides were identified as useful intermediates for the synthesis of the C17-C27 fragment of the bryostatins with control of the geometry of the exocyclic methoxycarbonylmethylene group. Following literature precedent, the Piers (E)-stereoselective addition of tributyltin hydride to an alkynoate followed by ester reduction and tin-iodine exchange gave vinylic iodides that could be used to form the C20-C21 bond of the bryostatins. Chelation controlled addition of lithiated 3-silyloxypropynes to 2-alkoxyaldehydes followed by reductive iodination was used to prepare vinylic iodides that could be used in the complementary assembly of the C21-C22 bond of the bryostatins. Initial studies of the synthesis of intermediates for metathesis studies using metal catalysed reactions of a vinylic iodide for C21-C22 bond formation were complicated by cyclisation reactions.Download high-res image (152KB)Download full-size image

Preliminary studies into the use of ring-closing metathesis (RCM) in a convergent approach for the total synthesis of bryostatins are described. An ester that would have provided an advanced intermediate for a synthesis of a 20-deoxybryostatin by a RCM was prepared from an unsaturated acid and alcohol corresponding to the C1–C16 and C17–C27 fragments. However, studies of the formation of the C16–C17 double-bond by RCM were not successful and complex mixtures of products were obtained. To provide an insight into factors that may be involved in hindering RCM in this system, a slightly simplified C1–C16 acid and modified C17–C25 alcohols were prepared and their use for the synthesis of analogues of bryostatins was investigated. Although only low yields were obtained, it appeared that macrolides analogous to the bryostatins can be prepared by RCM, using the Grubbs II catalyst, if the precursors lack the two methyl groups at C18. RCM was not observed, however, for substrates in which these methyl groups were present.

A series of 1′-(6-aminopurin-9-yl)-1′-deoxy-N-methyl-β-D-ribofuranuronamides that were characterised by 2-dialkylamino-7-methyloxazolo[4,5-b]pyridin-5-ylmethyl substituents on N6 of interest for screening as selective adenosine A3 receptor agonists, have been synthesised. This work involved the synthesis of 2-dialkylamino-5-aminomethyl-7-methyloxazolo[4,5-b]pyridines and analogues that were coupled with the known 1′-(6-chloropurin-9-yl)-1′-deoxy-N-methyl-β-D-ribofuranuronamide. The oxazolo[4,5-b]pyridines were synthesized by regioselective functionalisation of 2,4-dimethylpyridine N-oxides. The regioselectivities of these reactions were found to depend upon the nature of the heterocycle with 2-dimethylamino-5,7-dimethyloxazolo[4,5-b]pyridine-N-oxide undergoing regioselective functionalisation at the 7-methyl group on reaction with trifluoroacetic anhydride in contrast to the reaction of 4,6-dimethyl-3-hydroxypyridine-N-oxide with acetic anhydride that resulted in functionalisation of the 6-methyl group. To optimise selectivity for the A3 receptor, 5-aminomethyl-7-bromo-2-dimethylamino-4-[(3-methylisoxazol-5-yl)methoxy]benzo[d]oxazole was synthesised and coupled with the 1′-(6-chloropurin-9-yl)-1′-deoxy-N-methyl-β-D-ribofuranuronamide. The products were active as selective adenosine A3 agonists.

Syntheses of (1RS,2SR,6SR)-2-alkoxymethyl-, 2-hetaryl-, and 2-(hetarylmethyl)-7-arylmethyl-4,7-diaza-9-oxabicyclo[4.3.0]nonan-8-ones, of interest as potential muscarinic M1 receptor agonists, are described. A key step in the synthesis of (1RS,2SR,6SR)-7-benzyl-6-cyclobutyl-2-methoxymethyl-4,7-diaza-9-oxabicyclo[4.3.0]nonan-8-one, was the addition of isopropenylmagnesium bromide to 2-benzyloxycarbonylamino-3-tert-butyldimethylsilyloxy-2-cyclobutylpropanal. This gave the 4-tert-butyldimethylsilyloxymethyl-4-cyclobutyl-5-isopropenyloxazolidinone with the 5-isopropenyl and 4-tert-butyldimethylsilyloxymethyl groups cis-disposed about the five-membered ring by chelation controlled addition and in situ cyclisation. This reaction was useful for a range of organometallic reagents. The hydroboration–oxidation of (4SR,5RS)-3-benzyl-4-(tert-butyldimethylsilyloxymethyl)-4-cyclobutyl-5-(1-methoxyprop-2-en-2-yl)-1,3-oxazolidin-2-one gave (4SR,5RS)-3-benzyl-4-(tert-butyldimethylsilyloxymethyl)-4-cyclobutyl-5-[(SR)-1-hydroxy-3-methoxyprop-2-yl]-1,3-oxazolidin-2-one stereoselectively. 4,7-Diaza-9-oxabicyclo[4.3.0]nonan-8-ones with substituents at C2 that could facilitate C2 deprotonation were unstable with respect to oxazolidinone ring-opening and this restricted both the synthetic approach and choice of 2-heteroaryl substituent. The bicyclic system with a 2-furyl substituent at C2 was therefore identified as an important target. The addition of 1-lithio-1-(2-furyl)ethene to 2-benzyloxycarbonylamino-3-tert-butyldimethylsilyloxy-2-cyclobutylpropanal gave (4SR,5RS)-4-tert-butyldimethylsilyloxymethyl-4-cyclobutyl-5-[1-(2-furyl)ethenyl]-1,3-oxazolidinone after chelation controlled addition and in situ cyclisation. Following oxazolidinone N-benzylation, hydroboration at 35 °C, since hydroboration at 0 °C was unexpectedly selective for the undesired isomer, followed by oxidation gave a mixture of side-chain epimeric alcohols that were separated after SEM-protection and selective desilylation. Conversion of the neopentylic alcohols into the corresponding primary amines by reductive amination, was followed by N-nosylation, removal of the SEM-groups and cyclisation using a Mitsunobu reaction. Denosylation then gave the 2-furyloxazolidinonyl-fused piperidines, the (1RS,2SR,6SR)-epimer showing an allosteric agonistic effect on M1 receptors. Further studies resulted in the synthesis of other 2-substituted 4,7-diaza-9-oxabicyclo[4.3.0]nonan-8-ones and an analogous tetrahydropyran.

Synthetic approaches to (1RS,2SR,6SR)-7-arylmethyl-2-alkoxymethyl-4,7-diaza-9-oxabicyclo[4.3.0]nonan-8-ones, potentially selective muscarinic M1 receptor agonists, by hydration of 1,2,5,6-tetrahydropyridines were investigated. 3-Substituted N-tosyl-1,2,5,6-tetrahydropyridines were prepared by ring-closing metathesis (RCM). The direct hydration of these by hydroboration-oxidation was not usefully selective, but cis-3-hydroxymethyl-4-tert-butyldimethylsilyloxy-N-tosylpiperidine was prepared from 3-hydroxymethyl-N-tosyl-1,2,5,6-tetrahydropyridine by epoxidation, mesylation, reductive elimination, silylation and hydroboration-oxidation. Problems were encountered during attempts to prepare 3-alkoxymethyl-1,2,5,6-tetrahydropyridines with protected amino and cyclobutyl substituents at C5 by ring-closing metathesis, perhaps because of steric hindrance. Nevertheless interesting chemistry was encountered during the synthesis of the RCM precursors including a novel coupling via a 2-ethenyl-N-nosylaziridine and the formation of an oxaazathiocin by an intramolecular substitution of the nitro group of an N-nosyl protected amine by a proximate hydroxyl substituent.

•Results of interest in the context of the biosynthesis of phomactin A are described.•Advanced chemistry for the synthesis of phomactin A is reported.•Oxidation of homoallylic alcohols by TPAP gives unsaturated ketoaldehydes.•Reduction of a bicyclo[9.3.1]pentadecadien-14-one is affected by a C10 sulphone.•A dihydroxyepoxide is isolated rather than the isomeric hydroxytetrahydropyran.A dihydroxyepoxide analogous to that proposed as a late intermediate in the biosynthesis of phomactin A was prepared by reduction of the corresponding ketoaldehyde. The structure of the dihydroxyepoxide, specifically its configuration at C14, was confirmed by the X-ray crystal structure of its diacetate. The stereoselectivity of reduction of the ketoaldehyde would appear to be influenced by the presence of a remote phenylsulphonyl moiety at C10. Of interest in the context of the biosynthesis of phomactin A was the observation that this dihydroxyepoxide did not rearrange into the isomeric hydroxytetrahydropyran despite having the configuration at C14 required for this rearrangement.

Difluoroallylation of aldehydes mediated by indium powder has been carried out using 2-bromomethyl-1,1-difluorodec-1-ene leading to 2,2-difluoro-3-methylenealkanols in useful yields. The reaction has been applied to a range of aldehydes including both aromatic and aliphatic aldehydes.

2-Hydroxytetrahydropyrans corresponding to the C16–C27 fragment of bryostatins which have been difluorinated at C20 (bryostatin numbering) have been synthesised. The fluorine substituents were introduced by difluoroallylation. An (E)-selective Wittig reaction using a stabilised ylide provided the required methoxycarbonylmethylene substituent with excellent stereoselectivity.

On treatment with n-butyllithium, 4-alkoxy- and 4-silyloxy-2-(tributylstannylmethoxymethyl)-1,6-dimethyl-1-(phenylsulfonylmethyl)cyclohex-2-enes undergo tin–lithium exchange followed by [2,3]-Wittig rearrangements to give 3-alkoxy- and 3-silyloxy-2-hydroxymethyl-5,6-dimethyl-1-methylene-6-(phenylsulfonylmethyl)cyclohexanes in which the 2-hydroxymethyl and 6-phenylsulfonylmethyl residues are cis-disposed about the six-membered ring. In contrast, the corresponding 1-(phenylsulfanylmethyl)cyclohexenes give mainly methylenecyclohexanes with the 2-hydroxymethyl and 6-phenylsulfanylmethyl groups trans-disposed about the six-membered ring. This stereoselectivity is independent of the nature of the alkoxy- or silyloxy-substituent and configuration at C4. The 3-tert-butyldiphenylsilyloxy-1-methylene-6-(phenylsulfonylmethyl)cyclohexane was converted into a macrocyclic precursor of the phomactins.

The cis-fused lactones were the major products isolated from Diels–Alder reactions of (2E,4E)-2,4-dimethylhexa-2,4-dienyl methyl fumarate and maleate and from the cyclisation of the all (E)-2,4,6,8-tetramethyldeca-2,4,6,8-tetraenyl methyl fumarate in contrast to the Diels–Alder reactions of analogous substrates that lack the dienyl 2-methyl group. All of these Diels–Alder reactions led to the introduction of a methyl bearing quaternary centre. An intramolecular Diels–Alder reaction of a 3-(5,7-dimethylnona-5,7-dienoyl)pyrrolinone also gave mainly the endo-product, in this case with two adjacent quaternary centres.

Chaetochalasin A is a complex natural product whose biosynthesis may involve two domino Diels–Alder reactions. Approaches to the total synthesis of chaetochalasin A using this approach have been studied. Methyl (6R,8S,2Z,4E,10E,12E,14E)-6,8,10,14-tetramethylhexadeca-2,4,10,12,14-pentaenoate was identified as a key intermediate and was synthesized from (E)-1-bromo-4-tert-butyldimethylsilyloxy-2-methylbut-2-ene using diastereoselective alkylations of derivatives of (+)-pseudoephedrine to introduce the stereogenic centres, a modified Julia reaction to prepare the conjugated triene and a phosphonate condensation to provide the (2Z)-alkene. However, during the synthesis, facile geometrical isomerisation of the (14E)-trisubstituted and (2Z)-double-bonds was observed and attempts to incorporate this pentaene into a synthesis of chaetochalasin A led to the formation of mixtures of products. The analogous ethyl 6,8,10,14-tetramethylhexadeca-4,10,12,14-tetraenoate [that lacks the (2Z)-double-bond] was incorporated into a Diels–Alder precursor by acylation of a valine-derived N-acylpyrrolidinone followed by oxidative elimination of the corresponding 3-(phenylselanyl)pyrrolidinone. However, preliminary studies of the macrocycle-forming Diels–Alder reaction for a synthesis of chaetochalasin A were complicated by (E,Z)-isomerisation of the (10E)-double-bond of the conjugated triene and three Diels–Alder adducts were isolated and characterised. Further studies of this approach to chaetochalasin A will require an alternative procedure for the generation of the acylpyrrolinone in the presence of the acid sensitive conjugated triene.

A systematic process is introduced to compare 13C NMR spectra of two (or more) candidate samples of known structure to a natural product sample of unknown structure. The process is designed for the case where the spectra involved can reasonably be expected to be very similar, perhaps even identical. It is first validated by using published 13C NMR data sets for the natural product 4,6,8,10,16,18-hexamethyldocosane. Then the stereoselective total syntheses of two candidate isomers of the related 4,6,8,10,16-pentamethyldocosane natural product are described, and the process is applied to confidently assign the configuration of the natural product as (4S,6R,8R,10S,16S). This is accomplished even though the chemical shift differences between this isomer and its (16R)-epimer are only ±5–10 ppb (±0.005–0.01 ppm).

Reactions of 5-benzyloxy-4-methylpent-2-enyl(tributyl)stannane with aldehydes promoted by bismuth(III) iodide were usefully stereoselective in favour of the (E)-1,5-anti-6-benzyloxy-5-methylalk-3-en-1-ols. Similar stereoselectivity was observed for reactions of analogous 5-benzyloxy-4-methylpent-2-enyl bromides with aldehydes when promoted by a low valency bismuth species prepared by reduction of bismuth(III) triiodide with powdered zinc so providing a “tin-free” procedure. The analogous reactions of 4-benzyloxypent-2-enyl(tributyl)stannane with aldehydes promoted by bismuth(III) iodide were also stereoselective but gave lower yields. Attempted 1,6-stereocontrol using these reactions resulted in only modest stereoselectivities. Aspects of the chemistry of the products were studied in particular their stereoselective conversion into aliphatic compounds with methyl bearing stereogenic centres at 1,5,9,13- and 1,3,5-positions along the aliphatic chain. Mechanistically, allylic organobismuth species may be involved in both sets of reactions but this was not confirmed although the similar stereoselectivities observed for both the bismuth(III) iodide mediated reactions of the pent-2-enylstannanes and the low-valency bismuth promoted reactions of the pent-2-enyl bromides are consistent with participation of similar intermediates.

A procedure is reported for the introduction of both the thiazoline and (E)-dehydrobutyrine components into a tetrapeptide-derived fragment of vioprolides B and D. The (E)-dehydrobutyrine is introduced first but, as the carbon–carbon double-bond of the dehydrobutyrine appeared incompatible with an adjacent thiol, the thiazoline was assembled by dehydration of a serine containing thioamide not by dehydration of a cysteinyl analogue.

On treatment with acid, an open-chain 5-acylamino-3,8-diketo-ester, methyl (4R,5S,7S)-7-benzyloxy-4-[(S)-1-benzyloxyprop-2-yl]-5-methoxycarbonylamino-3,8-dioxododecanoate, cyclised via a stereoselective Mannich reaction to give an 8-azabicyclo[3.2.1]octanone. Hydrogenolysis of this with in situ acetal formation, reduction of the ester and a further cyclisation gave a lactam, (4R,5R,8S,9R,10S,12S,13S)-13-butyl-8-methyl-1-aza-6,14-dioxapentacyclo[8.3.0.04,1305,9.15,12]tetradecan-2-one, that corresponds to the pentacyclic core of stemofoline.An intramolecular Mannich reaction, hydrogenolysis and cyclisation led to the pentacyclic core of stemofoline.

Two syntheses of the C(7)–C(16)-fragment 41 of epothilone D 2 were developed that were based on tin(IV) bromide mediated reactions of 5,6-difunctionalised hex-2-enylstannanes with aldehydes. In the first synthesis, (5S)-6-tert-butyldimethylsilyloxy-5-hydroxy-2-methylhex-2-enyl(tributyl)stannane 20 was reacted with (E)-but-2-enal to give (2S,7R,4Z,8E)-1-tert-butyldimethylsilyloxy-5-methyldeca-4,8-diene-2,7-diol 26 containing ca. 20% of its (7S)-epimer. Following desilylation, the crystalline (2S,7R)-triol 32 was protected as its acetonide 33 and esterified to give the (4-methoxybenzyloxy)acetate 34. An Ireland–Claisen rearrangement of this ester gave methyl (2R,3S,10S,4E,7Z)-3,7-dimethyl-10,11-(dimethylmethylene)dioxy-2-(4-methoxybenzyloxy)undeca-4,7-dienoate 35 that was converted into (2S,9S,6Z)-2,6-dimethyl-9,10-(dimethylmethylene)dioxydec-6-en-1-ol 41 by regioselective alkene manipulation, ester reduction and cleavage of the resulting terminal diol 40 with a reductive work-up. The second synthesis involved the tin(IV) bromide mediated reaction between the stannane 20 and (3S)-4-(4-methoxybenzyloxy)-3-methylbutanal 44 that gave (2S,7S,9S,4Z)-1-tert-butyldimethylsilyloxy-5,9-dimethyl-10-(4-methoxybenzyloxy)dec-4-ene-2,7-diol 45 containing ca. 20% of its (7R)-epimer. After desilylation and protection of the vicinal diol as its acetonide 46, a Barton–McCombie reductive removal of the remaining hydroxyl group gave the (2S,9S,6Z)-2,6-dimethyl-9,10-(dimethylmethylene)dioxydec-6-en-1-ol 41 after oxidative removal of the PMB-ether. The first of these syntheses uses just one chiral starting material, but the second is shorter and more convergent. It was therefore modified by the use of (5S)-6-tert-butyldimethylsilyloxy-5-(2-trimethylsilylethoxy)methoxy-2-methylhex-2-enyl(tributyl)stannane 49 that reacted with (3S)-4-(4-methoxybenzyloxy)-3-methylbutanal 44 to give a 50:50 mixture of the C(4)-epimers of (2S,9S,6Z)-10-tert-butyldimethylsilyloxy-1-(4-methoxybenzyloxy)-2,6-dimethyl-9-(2-trimethylsilylethoxy)methoxydec-6-en-4-ol 50 with high fidelity for formation of the (Z)-alkene. Following the Barton–McCombie deoxygenation, the product 52 was taken through to (2S,9S,6Z,10E)-2,6,10-trimethyl-11-(2-methyl-1,3-thiazol-4-yl)-9-(2-trimethylsilylethoxy)methoxyundeca-6,10-dienal 59 that corresponded to the fully functionalised C(7)–C(17) fragment of epothilone D 2. A precedented stereoselective aldol condensation followed by O-protection, selective deprotection, oxidation and macrocyclisation then gave the macrolide 71 that was deprotected to complete a synthesis of epothilone D 2. Finally regio- and stereo-selective epoxidation gave epothilone B 1.

The tin(IV) bromide promoted reaction of 7-hydroxy-7-phenylhept-2-enyl(tributyl)stannane 11 with benzaldehyde gave a mixture of the epimeric 1,8-diphenyloct-3-ene-1,8-diols 12 and so indirect methods were developed for aliphatic 1,8-stereocontrol to complete diastereoselective syntheses of (±)-patulolide C 1 and (±)-epipatulolide C 40. (5Z)-3,7-syn-7-(2-Trimethylsilylethoxy)methoxyocta-1,5-dien-3-ol 17 was prepared from the tin(IV) chloride promoted reaction of 4-(2-trimethylsilylethoxy)methoxypent-2-enyl(tributyl)stannane 16 with acrolein (1,5-syn:1,5-anti = 96:4). An Ireland–Claisen rearrangement of the corresponding benzoyloxyacetate 21 with in situ esterification of the resulting acid using trimethylsilyldiazomethane gave methyl (4E,7Z)-2,9-anti-2-benzyloxy-9-(2-trimethylsilylethoxy)methoxydeca-4,7-dienoate 22 together with 10–15% of its 2,9-syn-epimer 26, the 2,9-syn-:2,9-anti-ratio depending on the conditions used. An 88:12 mixture of esters was taken through to the tert-butyldiphenylsilyl ether 38 of (±)-patulolide C 1 together with 6% of its epimer 39, by reduction, a Wittig homologation and deprotection/macrocyclisation. Following separation of the epimeric silyl ethers, deprotection of the major epimer 38 gave (±)-patulolide C 1. The success of 2,3-Wittig rearrangements of allyl ethers prepared from (5Z)-3,7-syn-7-(2-trimethylsilylethoxy)methoxyocta-1,5-dien-3-ol 17 was dependent on the substituents on the allyl ether. Best results were obtained using the pentadienyl ether 56 and the cinnamyl ether 49 that rearranged with >90:10 stereoselectivity in favour of (1E,5E,8Z)-3,10-syn-1-phenyl-10-(2-trimethylsilylethoxy)methoxyundeca-1,5,8-trien-3-ol 50. This product was taken through to the separable silyl ethers 38 and 39, ratio 7:93 by regioselective epoxidation and alkene reduction using diimide, followed by deoxygenation, ozonolysis, a Wittig homologation and selective deprotection/macrocyclisation. Deprotection of the major epimer 39 gave (±)-epipatulolide C 40.

The stereoselective reaction of an allyl bromide with an aldehyde mediated by a low valency bismuth species was the key reaction in stereoselective syntheses of (4S,6R,8R,10S,16S)- and (4S,6R,8R,10S,16R)-4,6,8,10,16-pentamethyldocosanes. 13C NMR data for these compounds confirmed that the cuticular hydrocarbon isolated from the cane beetle Antitrogus parvulus was the (4S,6R,8R,10S,16S)-stereoisomer.

The tin(IV) chloride mediated cyclisation of (Z)-homoallylic alcohols using phenylselenenyl chloride or phthalimide in the presence of a Lewis acid followed by reductive removal of the phenylselenenyl group was found to give 2,5-cis-disubstituted tetrahydrofurans with excellent stereocontrol. Using this procedure, (2S,4S,8R,6Z)-9-benzyloxy-2-tert-butyldiphenylsilyloxy-8-methylnon-6-en-4-ol (11), prepared stereoselectively via the tin(IV) chloride promoted reaction between the (R)-5-benzyloxy-4-methylpent-2-enyl(tributyl)stannane (3) and (S)-3-tert-butyldiphenylsilyloxybutanal (10), gave (2S,3R,6S,8S)-1-benzyloxy-8-tert-butyldiphenylsilyloxy-3,6-epoxy-2-methylnonane (13) after deselenation. This tetrahydrofuran was selectively deprotected, oxidized and esterified to give methyl nonactate (2). Having established this synthesis of 2,5-cis-disubstituted tetrahydrofurans, it was applied to complete a synthesis of pamamycin 607 (1). (2S,3R,6S,8R)-1-Benzyloxy-8-[N-methyl-N-(toluene-4-sulfonyl)amino]-3,6-epoxy-2-methylundecane (35) was prepared stereoselectively from (R)-3-[N-(toluene-4-sulfonyl)-N-methylamino]hexanal (32) by reaction with the stannane 3 followed by cyclisation of the resulting alkenol 33 and deselenation. Following debenzylation and oxidation, an aldol reaction of the aldehyde 37 using the lithium enolate of 2,6-dimethylphenyl propanoate (61) gave mainly the 2,3-anti-3,4-syn-adduct 48. After protection of the secondary alcohol as its tert-butyldimethylsilyl ether 49, reduction using DIBAL-H and oxidation, the resulting aldehyde, (2S,3S,4R,5R,8S,10R)-3-tert-butyldimethylsilyloxy-2,4-dimethyl-5,8-epoxy-10-[N-methyl-N-(toluene-4-sulfonyl)amino]tridecanal (62), was taken through to the bis-tetrahydrofuran 65 by repeating the sequence of the reactions with the stannane 3, cyclisation and deselenation. The N-(toluene-4-sulfonyl) group was then replaced by an N-(tert-butoxycarbonyl) group and O-debenzylation and oxidation gave the carboxylic acid 70 that corresponds to the C(1)–C(18) fragment of pamamycin 607 (1). Similar chemistry was used to prepare the C(1′)–C(11′) fragment 89 of the pamamycin, except that in this case the configuration of the secondary alcohol introduced by the allylstannane reaction had to be inverted using a Mitsunobu reaction before the cyclisation. Esterification of the carboxylic acid of the C(1)–C(18)-fragment 70 using the alcohol 89 of the C(1′)–C(11′) fragment followed by selective deprotection, macrocyclisation, N-deprotection and N-methylation gave pamamycin 607 (1) that was identical to a sample of the natural product.

Transmetallation of the 5-benzyloxy-4-methylpent-2-en-1-yl(tributyl)- and -(triphenyl)stannanes 1 and 8 using tin(IV) chloride generates an allyltin trichloride that reacts with aldehydes to give (Z)-1,5-anti-6-benzyloxy-5-methylhex-3-en-1-ols 2. The allyltin trichloride believed to be the key intermediate in these reactions has been trapped by phenyllithium to give anti-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 9. Transmetallation of this anti-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 9 generated an allyltin trichloride that reacted with aldehydes to give the (Z)-1,5-syn-6-benzyloxy-5-methylhex-3-en-1-ols 23 and was trapped by phenyllithium to give syn-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 24. Similar stereoselectivity was observed for tin(IV) chloride promoted reactions of this syn-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 24 with aldehydes and with phenyllithium. The allyltin trichlorides generated by transmetallation of 4-hydroxy- and 4-benzyloxy-pent-2-enyl(triphenyl)stannanes 34 and 35 were similarly trapped by phenyllithium to give 4-hydroxy- and 4-benzyloxy-pent-1-en-3-ylstannanes 36 and 37 whose configurations were established by correlation with known compounds. This work confirmed the configurations of the intermediate allyltin trichlorides involved in tin(IV) chloride promoted reactions of 4- and 5-alkoxypent-2-enylstannanes with aldehydes and showed that the high levels of remote stereocontrol were due mainly to kinetically controlled transmetallation. A fuller mechanistic scheme is proposed for the reactions in the 5-benzyloxy-4-methylpent-2-enylstannane series together with relevant 119Sn NMR data.

Reactions of the allyltin trichloride 45 generated from (4S)-4-benzyloxypent-2-enyl(tributyl)stannane 1 with imines prepared from glyoxylates proceed with useful levels of 1,5-stereocontrol in favour of (4E)-2,6-anti-2-(alkylamino)-6-benzyloxyhept-4-enoates 49. This stereoselectivity, controlled by the chirality of the stannane, dominates over any intrinsic stereochemical bias of the imine although a small amount of matching and mis-matching was observed. The allyltin trichloride 77 prepared from (4S)-4-(tert-butyldimethylsilyloxy)pent-2-enyl(tributyl)stannane 52 reacts with 1-alkoxycarbonylimines with the opposite 1,5-stereoselectivity to give the (4E)-2,6-syn-diastereoisomers 79. Matching and mismatching was more pronounced for tin(IV) chloride mediated reactions of (4R)-5-benzyloxy-4-methylpent-2-enyl(tributyl)stannane 80 with chiral 1-alkoxycarbonylimines but useful stereoselectivity in favour of (4E)-2,6-syn-2-alkyl- and arylthio-amino-7-benzyloxy-6-methylhept-4-enoates 177 was observed for reactions with achiral imines and similar, but reduced, stereoselectivity was observed for the 5-tert-butyldimethylsilyloxypentenylstannane 82. However, excellent 1,5-stereocontrol in favour of the (4E)-2,6-anti-isomers 179 was found using the 4,5-bis-alkoxypent-2-enylstannane 106. Modest (4E)-2,7-anti-stereoselectivity was observed in the analogous tin(IV) bromide mediated reactions of (S)-5-methoxy- and (S)-5-hydroxyhex-2-enyl(tributyl)stannanes (SSSSSSSSSSSS)-123 and (SSSSSSSSSSSS)-122 with achiral 1-alkoxycarbonylimines but in this series the intrinsic stereochemical bias of the imine controls the facial selectivity of reactions of chiral 1-alkoxycarbonylimines. Useful (4E)-2,6-anti-stereoselectivity was also observed in the tin(IV) chloride promoted reaction of the 4-benzyloxypent-2-enylstannane 1 with an oxime O-benzyl ether.

The 20-deoxybryostatin 40 has been prepared using a modified Julia olefination to form the 16,17-double-bond, followed by macrolactonisation, selective deprotection and oxidation. This is the first total synthesis of a 20-deoxybryostatin.

Alk-2-enylstannanes with 4-, 5- and 6-alkoxy- or -hydroxy-substituents are transmetallated stereoselectively with tin(IV) halides to generate allyltin trihalides which react with aldehydes to give (Z)-alk-3-enols with useful levels of 1,5-, 1,6- and 1,7-stereocontrol. Alk-2-enylstannanes with a stereogenic centre bearing a hydroxy or alkoxy group at the 4-, 5- or 6-position, react with overall (Z)-1,5-, 1,6- and 1,7-syn-stereoselectivity with respect to the hydroxy and alkoxy substituents. The analogous reactions of alkoxy- and -hydroxyalk-2-enylstannanes with a methyl bearing stereogenic centre at the 4- or 5-position react with overall (Z)-1,5- and 1,6-anti-stereoselectivity with respect to the hydroxy and methyl substituents.

Progress on a total synthesis of the marine natural products, the bryostatins, is reviewed. Following studies aimed at the synthesis of the 1,16- and 17,27-fragments, procedures for the assembly of the macrocyclic ring of the bryostatins were investigated. Although ring-closing metathesis was not found to be useful for the synthesis of bryostatins with geminal dimethyl groups at C18, the modified Julia reaction was found to be useful for the stereoselective formation of the 16,17-double-bond and led to a synthesis of an advanced macrocyclic intermediate. Several novel synthetic procedures feature in this work.

The combination of a [2,3]-Wittig rearrangement of a suitably substituted cyclohexenylmethyl propargyl ether with a subsequent conversion of the alkyne to a trisubstituted alkene and cyclisation via intramolecular sulfone alkylation has proved to be a useful stereoselective approach to advanced macrocyclic intermediates for a projected synthesis of phomactins. Thus Luche reduction of methyl (1RS,6SR)-2-(bromomethyl)-1,6-dimethyl-4-oxocyclohex-2-ene-1-carboxylate 24 gave methyl (1RS,4RS,6SR)-2-bromomethyl-4-hydroxy-1,6-dimethylcyclohex-2-ene-1-carboxylate 26 which was protected as its (2-trimethylsilylethoxy)methyl ether 27. O-Alkylation of (E)-8-tert-butyldiphenylsilyloxy-7-methyloct-6-en-2-yn-1-ol 17 using this bromide gave the corresponding ether 28. This was reduced and the resulting primary alcohol 29 converted into a phenylsulfonyl group by displacement of the corresponding mesylate by sodium thiophenoxide followed by oxidation. A [2,3]-Wittig rearrangement of the resulting propargylic ether 31 was stereoselective and gave predominantly (2RS,3SR,5RS,6SR)-2-[(1RS,6E)-8-tert-butyldiphenylsilyloxy-1-hydroxyoct-6-en-2-yn-1-yl)]-5,6-dimethyl-6-(phenylsulfonyl)methyl-3-(trimethylsilylethoxy)methoxy-1-methylenecyclohexane 37 together with its epimer at C(1′)38. Following protection as its 4-methoxybenzyl ether 39 with O-desilylation and conversion of the primary alcohol 40 into the corresponding bromide 41, cyclisation by intramolecular allylation of the sulfone gave (1SR,2RS,11SR,12RS,14SR,7E)-10-phenylsulfonyl-8,11,12-trimethyl-15-methylene-2-(4-methoxybenzyl)-14-(2-trimethylsilylethoxy)methoxybicyclo[9.3.1]pentadec-7-en-3-yne 42 and reductive desulfonylation and O-deprotection gave (1RS,2RS,11SR,12RS,14SR,7E)-8,11,12-trimethyl-15-methylene-14-(2-trimethylsilylethoxy)methoxybicyclo[9.3.1]pentadec-7-en-3-yn-2-ol 44. Analogous chemistry was carried out following protection of the Wittig rearrangement product as its tri-isopropylsilylether 45. To prepare the corresponding (3E,7E)-3,7-dienol, the Wittig rearrangement products 37 and 38 were oxidised to the corresponding ketone 54. Conjugate addition of thiophenol followed by substitution of the major phenylthio adduct 56 using lithium dimethylcuprate gave the corresponding (E)-conjugated enone 57 which was reduced using sodium borohydride and the resulting alcohol 58 converted into its benzyloxymethoxy ether 59. This was taken through to give (1RS,2RS,11SR,12RS,14SR,3E,6E)-4,8,11,12-tetramethyl-15-methylene-14-(2-trimethylsilylethoxy)methoxybicyclo[9.3.1]pentadeca-3,7-dien-2-ol 63 which has the full carbon skeleton of the phomactins.

Two approaches to tetrahydro-[1H]-2-benzazepin-4-ones of interest as potentially selective, muscarinic (M3) receptor antagonists have been developed. Base promoted addition of 2-(tert-butoxycarbonylamino)methyl-1,3-dithiane 5 with 2-(tert-butyldimethylsiloxymethyl)benzyl chloride 14 gave the corresponding 2,2-dialkylated 1,3-dithiane 15 which was taken through to the dithiane derivative 19 of the parent 2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-one by desilylation, oxidation and cyclisation via a reductive amination. After conversion into the N-tert-butyloxycarbonyl, N-toluene p-sulfonyl and N-benzyl derivatives 20–22, hydrolysis of the dithiane gave the N-protected tetrahydro-[1H]-2-benzazepin-4-ones 23–25. However, preliminary attempts to convert these into 5-cycloalkyl-5-hydroxy derivatives were not successful. In the second approach, ring-closing metathesis was used to prepare 2,3-dihydro-[1H]-2-benzazepines which were hydroxylated and oxidized to give the required 5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-ones. Following preliminary studies, ring-closing metathesis of the dienyl N-(2-nitrophenyl)sulfonamide 48 gave the dihydrobenzazepine 50 which was converted into the 2-butyl-5-cyclobutyl-5-hydroxytetrahydrobenzazepin-4-one 55 by hydroxylation and N-deprotection followed by N-alkylation via reductive amination, and oxidation. This chemistry was then used to prepare the 2-[(N-arylmethyl)aminoalkyl analogues 69, 72, 76 and 78. N-Acylation followed by amide reduction using the borane–tetrahydrofuran complex was also used to achieve N-alkylation of dihydrobenzazepines and this approach was used to prepare the 5-cyclopentyl-5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-one 103 and the 5-cyclobutyl-8-fluoro-5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-one 126. The structures of 2-tert-butyloxycarbonyl-4,4-propylenedithio-2,3,4,5-tetrahydro-[1H]-2-benzazepine 20 and (4RS,5SR)-2-butyl-5-cyclobutyl-4,5-dihydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepine 53 were confirmed by X-ray diffraction. The racemic 5-cycloalkyl-5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-ones were screened for muscarinic receptor antagonism. For M3 receptors from guinea pig ileum, these compounds had log10KB values of up to 7.2 with selectivities over M2 receptors from guinea pig left atria of approximately 40.

A synthesis of the racemic 6-aryloxymethyl-5-hydroxy-2,3,4,5-[1H]-2-tetrahydrobenzazepin-4-one 2, for evaluation as a muscarinic (M3) antagonist, is described. 2-[2-tert-Butyldimethylsilyloxymethyl-6-(2,6-dimethoxyphenoxymethyl)phenyl]propan-2-ol 10 was prepared from 2,6-dimethyl-1-bromobenzene 5 and taken through to N-[3-(2,6-dimethoxyphenoxymethyl)-2-(propen-2-yl)phenyl]methyl-N-prop-2-enyl 2-nitrobenzene sulfonamide 4. However, attempts to cyclise this diene by alkene metathesis were unsuccessful, the open-chain alkene 15 being the only product isolated in yields of up to 70%. In a second approach to the 6-aryloxymethyl-5-hydroxytetrahydrobenzazepin-4-one, methyl (Z)-3-[2-(1-tert-butyldimethylsilyloxymethyl)-6-(1,6-dimethoxyphenoxymethyl)phenyl]but-2-enoate 24 was converted into (Z)-3-[2-hydroxymethyl-6-(2,6-dimethoxyphenoxymethyl)phenyl]but-2-enyl 2-nitrobenzene sulfonamide 17 which was cyclised under Mitsunobu conditions to the corresponding 2,3-dihydro-[1H]-2-benzazepine 3. The structure of this was confirmed by an X-ray crystal structure of its 2-(4-bromophenylsulfonyl) analogue 28, and functional group modification including hydroxylation, attachment of the requisite side-chain at C(2) and further oxidation gave the target compound 2 which was assayed for muscarinic (M3) activity.

Preliminary studies of a synthetic approach to the alkaloid stemofoline1 are reported. Stereoselective cyclisation of the ketoester 14 gave the 1-butyl-2,8-bis(methoxycarbonyl)-8-azabicyclo[3.2.1]octane 21 in which the 2-methoxycarbonyl group is in the axial position. The analogous ketones 15, 18 and 19 were also cyclised to give the 8-azabicyclo[3.2.1]octanes 22–24 with axial electron-withdrawing 2-substituents. The structure of the bicyclic ketosulfone 22 was confirmed by X-ray diffraction. Conversion of ester 21 into the tricyclic lactams 31 and 39, in which the amide fragments are significantly distorted from planarity, was achieved by treatment of the iodides 29 and 38 with tert-butyllithium. The structure of the deprotected tricyclic hydroxylactam 40 was confirmed by X-ray diffraction, which showed the non-planar geometry of the lactam fragment and the distortion induced into the bicyclo[3.2.1]octane by the additional two-carbon bridge. This meant that the endo hydrogen at C9 was significantly closer to the 5-hydroxyl group than the endo hydrogen at C8. This structural feature was utilised to direct a regioselective remote oxidation of the hydroxylactam 40 using lead tetraacetate, which was accompanied by selective insertion into the closer endo C–H bond to give the tetracyclic ether 41. Lactam 39 was converted into the tricyclic aminoketone 49 by reduction to the aminol 44 using lithium aluminium hydride and reduction of the intermediate, possibly the chloride 46, formed from aminol 44 using thionyl chloride, with more lithium aluminium hydride, followed by O-deprotection and oxidation. The bicyclic ketoester 21 was also protected as its ketal 50, which was taken through via the tricyclic lactam 54 into the ketoamine 49. Finally, allylation of the tricyclic lactam 42 and amine 49 gave the axial allylated products 60 and 58, but further elaboration for incorporation of C10 and C11 (of stemofoline) was not straightforward. Alkylation of the protected hydroxyketone 64, which was prepared from the bicyclic ketoester 21, gave the axial alkylated products 65 and 69, and the ketoester 69 was converted into the tricyclic hydroxylactone 73. However, the formation of a tetracyclic lactam by treatment of the iodide 75 with tert-butyllithium was not successful.

Following studies using benzyloxymethyl isopropenyl ketone 5 and ethyl 3-(3-furyl)-3-oxopropanoate 6, Robinson reactions between aryloxymethyl isopropenyl ketones 19 and 5 and ethyl 3-(2-trimethylsilyl-3-furyl)-3-oxopropanoate 20 were found to be stereoselective giving cyclohexanones 21 and 41, in which the 3-(arylmethoxy) substituents were cis to the 2-hydroxyl groups, as the major products. After reduction and protection of ketone 21, selective PMB-deprotection, oxidation and stereoselective reduction inverted the configuration at C(3) to give the diol 30. Protection of the secondary 3-hydroxyl group followed by modification of the protected 4-alcohol then gave the hydroxybutenolides 36 and 37 after oxidation of the silylated furan using singlet oxygen. The 3-benzyloxycyclohexanone 41 was also converted into the hydroxybutenolide 37via the (2-trimethylsilylethoxy)methyl (SEM) ether 35.The Wittig reaction between the ylid generated from 2-methylpropyl(triphenyl)phosphonium salt and hydroxybutenolide 36 gave predominantly the (2Z,4Z)-dienyl acid 38 which was taken through to the butenolide 40. Similarly, the racemic hydroxybutenolide 37 was condensed with the racemic ylid derived from phosphonium salt 53 to give, after SEM-deprotection and 5-membered lactone formation, a mixture of the (9Z,2′Z)-dienyl lactones 58 and 59 containing ca. 10% of the corresponding (9Z,2′E)-isomers 60 and 61. (2′Z)/(2′E)-Isomerisation of the dienes 58 and 59 using iodine followed by deprotection gave a mixture of the seco-acids 62 and 63. Selective macrocyclisation of the seco-acid 62 in which the relative configuration of the C(1)–C(7) and C(17)–C(19) fragments (milbemycin numbering) corresponded to that present in the natural milbemycins, gave the β-milbemycin analogue 65 after butenolide reduction. The hydroxybutenolide 37 was also condensed with the ylid derived from the phosphonium salt 1 and the product taken through to (6R)-6-hydroxy-3,4-dihydromilbemycin E 77.Preliminary attempts to convert the β-milbemycin analogues 65 and 77 into tetrahydrofurans corresponding to analogues of α-milbemycins by treatment with toluene p-sulfonyl chloride under basic conditions gave the primary allylic chlorides 78 and 79.

A synthesis of the hydroxybutenolide (−)-6 required for synthesis of α-milbemycins and the completion of a total synthesis of milbemycin G 7 is reported.Following preliminary studies, an optimised synthesis of the hydroxybutenolide (−)-6 from the hydroxyketone 38 was developed which involved the resolution of 38 by separation of the 3-(O-chloroacetyl)-(S)-mandelates 80 and 83. Ester 80, which corresponded to the required enantiomer of the hydroxyketone 38, crystallized from the mixture of the diastereoisomeric esters 80 and 83 giving the (−)-hydroxyketone (−)-38 in an overall yield of 47% (based on racemic 38) after ethanolysis. Hydroxyketone (−)-38 was oxidised to the enolic diketone (−)-39 and phenylselenation and stereoselective reduction gave the trihydroxycyclohexyl selenide (−)-43. The regioselective introduction of the non-conjugated double-bond into the six-membered ring was then achieved by esterification of the 4-hydroxyl group using trichloroacetic acid to give the trichloroacetate (−)-69. Oxidative elimination from the trichloroacetate using tert-butyl hydroperoxide was highly regioselective and gave the endo- and exocyclic alkenes (−)-44 and (−)-46 in a ratio of 95 : 5 after ethanolysis of the trichloroacetates. Selective O-methylation of the 4-hydroxyl group via the cyclic stannylene 55 and protection of the 3-hydroxyl group as its 2-trimethylsilylethoxymethyl (SEM) ether gave the ester (−)-57. Following saponification of the ethyl ester, re-esterification using 2-trimethylsilylethanol and oxidation of the 2-trimethylsilylfuryl fragment using singlet oxygen gave the required hydroxybutenolide (−)-6.The Wittig reaction between the phosphonium salt 2 and the hydroxybutenolide (−)-6 gave a ca. 2 : 1 mixture of the (4Z)- and (4E)-isomers of the ester 84 which on treatment with a catalytic amount of iodine was converted into the (4E)-isomer (4E)-84. Deprotection gave the seco-acid 85 but attempts to macrocyclise this were unsuccessful, the elimination product 86 being the only product isolated. The Wittig product 84 was taken through to the butenolide (2′E)-91 by removal of the SEM group, cyclisation to form the butenolide ring and diene isomerization, but this could not be converted into the corresponding seco-acid 92. However, removal of the SEM group from the seco-acid 85 gave the trihydroxy-acid 93 which was cyclized under modified Yamaguchi conditions to give the macrolide 94 together with a small amount of the macrocyclic butenolide 95. Reduction of this mixture using diisobutylaluminium hydride gave (6R)-6-hydroxymilbemycin E 96 which was converted to milbemycin G 7 by cyclisation of the primary chloride 97. The synthetic milbemycin G 7 was identical to a sample prepared by methylation of a commercial sample of milbemycin D 98, 7-O-methylmilbemycin G 99 being a side-product of this methylation.

A synthesis of the racemic 6-aryloxymethyl-5-hydroxy-2,3,4,5-[1H]-2-tetrahydrobenzazepin-4-one 2, for evaluation as a muscarinic (M3) antagonist, is described. 2-[2-tert-Butyldimethylsilyloxymethyl-6-(2,6-dimethoxyphenoxymethyl)phenyl]propan-2-ol 10 was prepared from 2,6-dimethyl-1-bromobenzene 5 and taken through to N-[3-(2,6-dimethoxyphenoxymethyl)-2-(propen-2-yl)phenyl]methyl-N-prop-2-enyl 2-nitrobenzene sulfonamide 4. However, attempts to cyclise this diene by alkene metathesis were unsuccessful, the open-chain alkene 15 being the only product isolated in yields of up to 70%. In a second approach to the 6-aryloxymethyl-5-hydroxytetrahydrobenzazepin-4-one, methyl (Z)-3-[2-(1-tert-butyldimethylsilyloxymethyl)-6-(1,6-dimethoxyphenoxymethyl)phenyl]but-2-enoate 24 was converted into (Z)-3-[2-hydroxymethyl-6-(2,6-dimethoxyphenoxymethyl)phenyl]but-2-enyl 2-nitrobenzene sulfonamide 17 which was cyclised under Mitsunobu conditions to the corresponding 2,3-dihydro-[1H]-2-benzazepine 3. The structure of this was confirmed by an X-ray crystal structure of its 2-(4-bromophenylsulfonyl) analogue 28, and functional group modification including hydroxylation, attachment of the requisite side-chain at C(2) and further oxidation gave the target compound 2 which was assayed for muscarinic (M3) activity.

The development of a synthesis of the C1–C16 fragment of bryostatins in which the key step is a stereoselective oxy-Michael reaction used to assemble the cis-2,6-disubstituted tetrahydropyran with the exocyclic alkene already installed, is described. Following early work using Stille reactions to prepare precursors for oxy-Michael reactions, a more efficient route was devised based on a ketophosphonate-aldehyde condensation to prepare the key dienone. Synthesis of the aldehyde required for the ketophosphonate condensation involved the highly stereoselective addition of a diorganocuprate derived from allylmagnesium bromide and copper(I) iodide to the methyl 5-hydroxyhex-2-ynoate prepared by ring-opening of a protected glycidol using lithiated methyl propiolate. Ester reduction, selective alcohol protection and oxidative cleavage of the terminal double bond gave the required aldehyde. The ketophosphonate was prepared in 13 steps from (R)-pantolactone using a synthesis based on a chelation controlled aldol condensation and an anti-selective reduction of a 3-hydroxyketone. Following the ketophosphonate-aldehyde reaction, selective deprotection followed by treatment with base effected the oxy-Michael reaction that gave the cis-2,6-disubstituted tetrahydropyran via thermodynamic control. Subsequent functional group manipulation led to the synthesis of a hydroxy ester that corresponded to the C1–C16 fragment of the bryostatins in 23 steps from (R)-pantolactone. The synthesis was repeated using slightly different protecting groups for a study of a ring-closing metathesis approach to the bryostatins.

Transmetallation of the 5-benzyloxy-4-methylpent-2-en-1-yl(tributyl)- and -(triphenyl)stannanes 1 and 8 using tin(IV) chloride generates an allyltin trichloride that reacts with aldehydes to give (Z)-1,5-anti-6-benzyloxy-5-methylhex-3-en-1-ols 2. The allyltin trichloride believed to be the key intermediate in these reactions has been trapped by phenyllithium to give anti-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 9. Transmetallation of this anti-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 9 generated an allyltin trichloride that reacted with aldehydes to give the (Z)-1,5-syn-6-benzyloxy-5-methylhex-3-en-1-ols 23 and was trapped by phenyllithium to give syn-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 24. Similar stereoselectivity was observed for tin(IV) chloride promoted reactions of this syn-5-benzyloxy-4-methylpent-1-en-3-yl(triphenyl)stannane 24 with aldehydes and with phenyllithium. The allyltin trichlorides generated by transmetallation of 4-hydroxy- and 4-benzyloxy-pent-2-enyl(triphenyl)stannanes 34 and 35 were similarly trapped by phenyllithium to give 4-hydroxy- and 4-benzyloxy-pent-1-en-3-ylstannanes 36 and 37 whose configurations were established by correlation with known compounds. This work confirmed the configurations of the intermediate allyltin trichlorides involved in tin(IV) chloride promoted reactions of 4- and 5-alkoxypent-2-enylstannanes with aldehydes and showed that the high levels of remote stereocontrol were due mainly to kinetically controlled transmetallation. A fuller mechanistic scheme is proposed for the reactions in the 5-benzyloxy-4-methylpent-2-enylstannane series together with relevant 119Sn NMR data.

The tin(IV) chloride mediated cyclisation of (Z)-homoallylic alcohols using phenylselenenyl chloride or phthalimide in the presence of a Lewis acid followed by reductive removal of the phenylselenenyl group was found to give 2,5-cis-disubstituted tetrahydrofurans with excellent stereocontrol. Using this procedure, (2S,4S,8R,6Z)-9-benzyloxy-2-tert-butyldiphenylsilyloxy-8-methylnon-6-en-4-ol (11), prepared stereoselectively via the tin(IV) chloride promoted reaction between the (R)-5-benzyloxy-4-methylpent-2-enyl(tributyl)stannane (3) and (S)-3-tert-butyldiphenylsilyloxybutanal (10), gave (2S,3R,6S,8S)-1-benzyloxy-8-tert-butyldiphenylsilyloxy-3,6-epoxy-2-methylnonane (13) after deselenation. This tetrahydrofuran was selectively deprotected, oxidized and esterified to give methyl nonactate (2). Having established this synthesis of 2,5-cis-disubstituted tetrahydrofurans, it was applied to complete a synthesis of pamamycin 607 (1). (2S,3R,6S,8R)-1-Benzyloxy-8-[N-methyl-N-(toluene-4-sulfonyl)amino]-3,6-epoxy-2-methylundecane (35) was prepared stereoselectively from (R)-3-[N-(toluene-4-sulfonyl)-N-methylamino]hexanal (32) by reaction with the stannane 3 followed by cyclisation of the resulting alkenol 33 and deselenation. Following debenzylation and oxidation, an aldol reaction of the aldehyde 37 using the lithium enolate of 2,6-dimethylphenyl propanoate (61) gave mainly the 2,3-anti-3,4-syn-adduct 48. After protection of the secondary alcohol as its tert-butyldimethylsilyl ether 49, reduction using DIBAL-H and oxidation, the resulting aldehyde, (2S,3S,4R,5R,8S,10R)-3-tert-butyldimethylsilyloxy-2,4-dimethyl-5,8-epoxy-10-[N-methyl-N-(toluene-4-sulfonyl)amino]tridecanal (62), was taken through to the bis-tetrahydrofuran 65 by repeating the sequence of the reactions with the stannane 3, cyclisation and deselenation. The N-(toluene-4-sulfonyl) group was then replaced by an N-(tert-butoxycarbonyl) group and O-debenzylation and oxidation gave the carboxylic acid 70 that corresponds to the C(1)–C(18) fragment of pamamycin 607 (1). Similar chemistry was used to prepare the C(1′)–C(11′) fragment 89 of the pamamycin, except that in this case the configuration of the secondary alcohol introduced by the allylstannane reaction had to be inverted using a Mitsunobu reaction before the cyclisation. Esterification of the carboxylic acid of the C(1)–C(18)-fragment 70 using the alcohol 89 of the C(1′)–C(11′) fragment followed by selective deprotection, macrocyclisation, N-deprotection and N-methylation gave pamamycin 607 (1) that was identical to a sample of the natural product.

The stereoselective reaction of an allyl bromide with an aldehyde mediated by a low valency bismuth species was the key reaction in stereoselective syntheses of (4S,6R,8R,10S,16S)- and (4S,6R,8R,10S,16R)-4,6,8,10,16-pentamethyldocosanes. 13C NMR data for these compounds confirmed that the cuticular hydrocarbon isolated from the cane beetle Antitrogus parvulus was the (4S,6R,8R,10S,16S)-stereoisomer.

Two approaches to the synthesis of compounds corresponding to the C17–C27 fragment of the 20-deoxybryostatins are described. The first approach is based on the palladium(0) catalysed coupling of tin enolates, generated in situ from enol acetates using tributyltin methoxide, with vinylic bromides. The vinylic bromides were prepared using the Sharpless asymmetric dihydroxylation to introduce the hydroxyl groups corresponding to those at C25 and C26 in the bryostatins. Following several steps to introduce alkynyl ester functionality, the stereoselective addition of a tributyltin cuprate followed by tributyltin–bromine exchange gave the required vinylic bromides. The palladium(0) catalysed couplings worked very well for enol esters containing thioether substituents and gave products with retention of the position and geometry of the trisubstituted double bond derived from the vinylic bromide. These were taken through to compounds corresponding to fully developed C17–C27 fragments ready for assembly of the 16,17-double-bond of bryostatins by Julia reactions. This chemistry was also applied to prepare intermediates suitable for incorporation into bryostatins by ring-closing metathesis but, in this case, the coupling reaction gave mixtures of products including both the required βγ-unsaturated ketone and a conjugated diene formed by a competing Heck reaction. To avoid this problem, a second approach to compounds suitable for incorporation into a metathesis-based assembly of 20-deoxybryostatins was developed. In this organotin-free synthesis, the key step was the conjugate addition of an organic cuprate generated from allylmagnesium bromide to an alkynoate that gave the required (Z)-trisubstituted alkene with excellent stereoselectivity. This was converted into metathesis precursors in a few steps.

Preliminary studies of a synthetic approach to the alkaloid stemofoline1 are reported. Stereoselective cyclisation of the ketoester 14 gave the 1-butyl-2,8-bis(methoxycarbonyl)-8-azabicyclo[3.2.1]octane 21 in which the 2-methoxycarbonyl group is in the axial position. The analogous ketones 15, 18 and 19 were also cyclised to give the 8-azabicyclo[3.2.1]octanes 22–24 with axial electron-withdrawing 2-substituents. The structure of the bicyclic ketosulfone 22 was confirmed by X-ray diffraction. Conversion of ester 21 into the tricyclic lactams 31 and 39, in which the amide fragments are significantly distorted from planarity, was achieved by treatment of the iodides 29 and 38 with tert-butyllithium. The structure of the deprotected tricyclic hydroxylactam 40 was confirmed by X-ray diffraction, which showed the non-planar geometry of the lactam fragment and the distortion induced into the bicyclo[3.2.1]octane by the additional two-carbon bridge. This meant that the endo hydrogen at C9 was significantly closer to the 5-hydroxyl group than the endo hydrogen at C8. This structural feature was utilised to direct a regioselective remote oxidation of the hydroxylactam 40 using lead tetraacetate, which was accompanied by selective insertion into the closer endo C–H bond to give the tetracyclic ether 41. Lactam 39 was converted into the tricyclic aminoketone 49 by reduction to the aminol 44 using lithium aluminium hydride and reduction of the intermediate, possibly the chloride 46, formed from aminol 44 using thionyl chloride, with more lithium aluminium hydride, followed by O-deprotection and oxidation. The bicyclic ketoester 21 was also protected as its ketal 50, which was taken through via the tricyclic lactam 54 into the ketoamine 49. Finally, allylation of the tricyclic lactam 42 and amine 49 gave the axial allylated products 60 and 58, but further elaboration for incorporation of C10 and C11 (of stemofoline) was not straightforward. Alkylation of the protected hydroxyketone 64, which was prepared from the bicyclic ketoester 21, gave the axial alkylated products 65 and 69, and the ketoester 69 was converted into the tricyclic hydroxylactone 73. However, the formation of a tetracyclic lactam by treatment of the iodide 75 with tert-butyllithium was not successful.

Two approaches to tetrahydro-[1H]-2-benzazepin-4-ones of interest as potentially selective, muscarinic (M3) receptor antagonists have been developed. Base promoted addition of 2-(tert-butoxycarbonylamino)methyl-1,3-dithiane 5 with 2-(tert-butyldimethylsiloxymethyl)benzyl chloride 14 gave the corresponding 2,2-dialkylated 1,3-dithiane 15 which was taken through to the dithiane derivative 19 of the parent 2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-one by desilylation, oxidation and cyclisation via a reductive amination. After conversion into the N-tert-butyloxycarbonyl, N-toluene p-sulfonyl and N-benzyl derivatives 20–22, hydrolysis of the dithiane gave the N-protected tetrahydro-[1H]-2-benzazepin-4-ones 23–25. However, preliminary attempts to convert these into 5-cycloalkyl-5-hydroxy derivatives were not successful. In the second approach, ring-closing metathesis was used to prepare 2,3-dihydro-[1H]-2-benzazepines which were hydroxylated and oxidized to give the required 5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-ones. Following preliminary studies, ring-closing metathesis of the dienyl N-(2-nitrophenyl)sulfonamide 48 gave the dihydrobenzazepine 50 which was converted into the 2-butyl-5-cyclobutyl-5-hydroxytetrahydrobenzazepin-4-one 55 by hydroxylation and N-deprotection followed by N-alkylation via reductive amination, and oxidation. This chemistry was then used to prepare the 2-[(N-arylmethyl)aminoalkyl analogues 69, 72, 76 and 78. N-Acylation followed by amide reduction using the borane–tetrahydrofuran complex was also used to achieve N-alkylation of dihydrobenzazepines and this approach was used to prepare the 5-cyclopentyl-5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-one 103 and the 5-cyclobutyl-8-fluoro-5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-one 126. The structures of 2-tert-butyloxycarbonyl-4,4-propylenedithio-2,3,4,5-tetrahydro-[1H]-2-benzazepine 20 and (4RS,5SR)-2-butyl-5-cyclobutyl-4,5-dihydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepine 53 were confirmed by X-ray diffraction. The racemic 5-cycloalkyl-5-hydroxy-2,3,4,5-tetrahydro-[1H]-2-benzazepin-4-ones were screened for muscarinic receptor antagonism. For M3 receptors from guinea pig ileum, these compounds had log10KB values of up to 7.2 with selectivities over M2 receptors from guinea pig left atria of approximately 40.

The combination of a [2,3]-Wittig rearrangement of a suitably substituted cyclohexenylmethyl propargyl ether with a subsequent conversion of the alkyne to a trisubstituted alkene and cyclisation via intramolecular sulfone alkylation has proved to be a useful stereoselective approach to advanced macrocyclic intermediates for a projected synthesis of phomactins. Thus Luche reduction of methyl (1RS,6SR)-2-(bromomethyl)-1,6-dimethyl-4-oxocyclohex-2-ene-1-carboxylate 24 gave methyl (1RS,4RS,6SR)-2-bromomethyl-4-hydroxy-1,6-dimethylcyclohex-2-ene-1-carboxylate 26 which was protected as its (2-trimethylsilylethoxy)methyl ether 27. O-Alkylation of (E)-8-tert-butyldiphenylsilyloxy-7-methyloct-6-en-2-yn-1-ol 17 using this bromide gave the corresponding ether 28. This was reduced and the resulting primary alcohol 29 converted into a phenylsulfonyl group by displacement of the corresponding mesylate by sodium thiophenoxide followed by oxidation. A [2,3]-Wittig rearrangement of the resulting propargylic ether 31 was stereoselective and gave predominantly (2RS,3SR,5RS,6SR)-2-[(1RS,6E)-8-tert-butyldiphenylsilyloxy-1-hydroxyoct-6-en-2-yn-1-yl)]-5,6-dimethyl-6-(phenylsulfonyl)methyl-3-(trimethylsilylethoxy)methoxy-1-methylenecyclohexane 37 together with its epimer at C(1′)38. Following protection as its 4-methoxybenzyl ether 39 with O-desilylation and conversion of the primary alcohol 40 into the corresponding bromide 41, cyclisation by intramolecular allylation of the sulfone gave (1SR,2RS,11SR,12RS,14SR,7E)-10-phenylsulfonyl-8,11,12-trimethyl-15-methylene-2-(4-methoxybenzyl)-14-(2-trimethylsilylethoxy)methoxybicyclo[9.3.1]pentadec-7-en-3-yne 42 and reductive desulfonylation and O-deprotection gave (1RS,2RS,11SR,12RS,14SR,7E)-8,11,12-trimethyl-15-methylene-14-(2-trimethylsilylethoxy)methoxybicyclo[9.3.1]pentadec-7-en-3-yn-2-ol 44. Analogous chemistry was carried out following protection of the Wittig rearrangement product as its tri-isopropylsilylether 45. To prepare the corresponding (3E,7E)-3,7-dienol, the Wittig rearrangement products 37 and 38 were oxidised to the corresponding ketone 54. Conjugate addition of thiophenol followed by substitution of the major phenylthio adduct 56 using lithium dimethylcuprate gave the corresponding (E)-conjugated enone 57 which was reduced using sodium borohydride and the resulting alcohol 58 converted into its benzyloxymethoxy ether 59. This was taken through to give (1RS,2RS,11SR,12RS,14SR,3E,6E)-4,8,11,12-tetramethyl-15-methylene-14-(2-trimethylsilylethoxy)methoxybicyclo[9.3.1]pentadeca-3,7-dien-2-ol 63 which has the full carbon skeleton of the phomactins.

Chaetochalasin A is a complex natural product whose biosynthesis may involve two domino Diels–Alder reactions. Approaches to the total synthesis of chaetochalasin A using this approach have been studied. Methyl (6R,8S,2Z,4E,10E,12E,14E)-6,8,10,14-tetramethylhexadeca-2,4,10,12,14-pentaenoate was identified as a key intermediate and was synthesized from (E)-1-bromo-4-tert-butyldimethylsilyloxy-2-methylbut-2-ene using diastereoselective alkylations of derivatives of (+)-pseudoephedrine to introduce the stereogenic centres, a modified Julia reaction to prepare the conjugated triene and a phosphonate condensation to provide the (2Z)-alkene. However, during the synthesis, facile geometrical isomerisation of the (14E)-trisubstituted and (2Z)-double-bonds was observed and attempts to incorporate this pentaene into a synthesis of chaetochalasin A led to the formation of mixtures of products. The analogous ethyl 6,8,10,14-tetramethylhexadeca-4,10,12,14-tetraenoate [that lacks the (2Z)-double-bond] was incorporated into a Diels–Alder precursor by acylation of a valine-derived N-acylpyrrolidinone followed by oxidative elimination of the corresponding 3-(phenylselanyl)pyrrolidinone. However, preliminary studies of the macrocycle-forming Diels–Alder reaction for a synthesis of chaetochalasin A were complicated by (E,Z)-isomerisation of the (10E)-double-bond of the conjugated triene and three Diels–Alder adducts were isolated and characterised. Further studies of this approach to chaetochalasin A will require an alternative procedure for the generation of the acylpyrrolinone in the presence of the acid sensitive conjugated triene.

Preliminary studies into the use of ring-closing metathesis (RCM) in a convergent approach for the total synthesis of bryostatins are described. An ester that would have provided an advanced intermediate for a synthesis of a 20-deoxybryostatin by a RCM was prepared from an unsaturated acid and alcohol corresponding to the C1–C16 and C17–C27 fragments. However, studies of the formation of the C16–C17 double-bond by RCM were not successful and complex mixtures of products were obtained. To provide an insight into factors that may be involved in hindering RCM in this system, a slightly simplified C1–C16 acid and modified C17–C25 alcohols were prepared and their use for the synthesis of analogues of bryostatins was investigated. Although only low yields were obtained, it appeared that macrolides analogous to the bryostatins can be prepared by RCM, using the Grubbs II catalyst, if the precursors lack the two methyl groups at C18. RCM was not observed, however, for substrates in which these methyl groups were present.

A series of 1′-(6-aminopurin-9-yl)-1′-deoxy-N-methyl-β-D-ribofuranuronamides that were characterised by 2-dialkylamino-7-methyloxazolo[4,5-b]pyridin-5-ylmethyl substituents on N6 of interest for screening as selective adenosine A3 receptor agonists, have been synthesised. This work involved the synthesis of 2-dialkylamino-5-aminomethyl-7-methyloxazolo[4,5-b]pyridines and analogues that were coupled with the known 1′-(6-chloropurin-9-yl)-1′-deoxy-N-methyl-β-D-ribofuranuronamide. The oxazolo[4,5-b]pyridines were synthesized by regioselective functionalisation of 2,4-dimethylpyridine N-oxides. The regioselectivities of these reactions were found to depend upon the nature of the heterocycle with 2-dimethylamino-5,7-dimethyloxazolo[4,5-b]pyridine-N-oxide undergoing regioselective functionalisation at the 7-methyl group on reaction with trifluoroacetic anhydride in contrast to the reaction of 4,6-dimethyl-3-hydroxypyridine-N-oxide with acetic anhydride that resulted in functionalisation of the 6-methyl group. To optimise selectivity for the A3 receptor, 5-aminomethyl-7-bromo-2-dimethylamino-4-[(3-methylisoxazol-5-yl)methoxy]benzo[d]oxazole was synthesised and coupled with the 1′-(6-chloropurin-9-yl)-1′-deoxy-N-methyl-β-D-ribofuranuronamide. The products were active as selective adenosine A3 agonists.

Syntheses of (1RS,2SR,6SR)-2-alkoxymethyl-, 2-hetaryl-, and 2-(hetarylmethyl)-7-arylmethyl-4,7-diaza-9-oxabicyclo[4.3.0]nonan-8-ones, of interest as potential muscarinic M1 receptor agonists, are described. A key step in the synthesis of (1RS,2SR,6SR)-7-benzyl-6-cyclobutyl-2-methoxymethyl-4,7-diaza-9-oxabicyclo[4.3.0]nonan-8-one, was the addition of isopropenylmagnesium bromide to 2-benzyloxycarbonylamino-3-tert-butyldimethylsilyloxy-2-cyclobutylpropanal. This gave the 4-tert-butyldimethylsilyloxymethyl-4-cyclobutyl-5-isopropenyloxazolidinone with the 5-isopropenyl and 4-tert-butyldimethylsilyloxymethyl groups cis-disposed about the five-membered ring by chelation controlled addition and in situ cyclisation. This reaction was useful for a range of organometallic reagents. The hydroboration–oxidation of (4SR,5RS)-3-benzyl-4-(tert-butyldimethylsilyloxymethyl)-4-cyclobutyl-5-(1-methoxyprop-2-en-2-yl)-1,3-oxazolidin-2-one gave (4SR,5RS)-3-benzyl-4-(tert-butyldimethylsilyloxymethyl)-4-cyclobutyl-5-[(SR)-1-hydroxy-3-methoxyprop-2-yl]-1,3-oxazolidin-2-one stereoselectively. 4,7-Diaza-9-oxabicyclo[4.3.0]nonan-8-ones with substituents at C2 that could facilitate C2 deprotonation were unstable with respect to oxazolidinone ring-opening and this restricted both the synthetic approach and choice of 2-heteroaryl substituent. The bicyclic system with a 2-furyl substituent at C2 was therefore identified as an important target. The addition of 1-lithio-1-(2-furyl)ethene to 2-benzyloxycarbonylamino-3-tert-butyldimethylsilyloxy-2-cyclobutylpropanal gave (4SR,5RS)-4-tert-butyldimethylsilyloxymethyl-4-cyclobutyl-5-[1-(2-furyl)ethenyl]-1,3-oxazolidinone after chelation controlled addition and in situ cyclisation. Following oxazolidinone N-benzylation, hydroboration at 35 °C, since hydroboration at 0 °C was unexpectedly selective for the undesired isomer, followed by oxidation gave a mixture of side-chain epimeric alcohols that were separated after SEM-protection and selective desilylation. Conversion of the neopentylic alcohols into the corresponding primary amines by reductive amination, was followed by N-nosylation, removal of the SEM-groups and cyclisation using a Mitsunobu reaction. Denosylation then gave the 2-furyloxazolidinonyl-fused piperidines, the (1RS,2SR,6SR)-epimer showing an allosteric agonistic effect on M1 receptors. Further studies resulted in the synthesis of other 2-substituted 4,7-diaza-9-oxabicyclo[4.3.0]nonan-8-ones and an analogous tetrahydropyran.

Alk-2-enylstannanes with 4-, 5- and 6-alkoxy- or -hydroxy-substituents are transmetallated stereoselectively with tin(IV) halides to generate allyltin trihalides which react with aldehydes to give (Z)-alk-3-enols with useful levels of 1,5-, 1,6- and 1,7-stereocontrol. Alk-2-enylstannanes with a stereogenic centre bearing a hydroxy or alkoxy group at the 4-, 5- or 6-position, react with overall (Z)-1,5-, 1,6- and 1,7-syn-stereoselectivity with respect to the hydroxy and alkoxy substituents. The analogous reactions of alkoxy- and -hydroxyalk-2-enylstannanes with a methyl bearing stereogenic centre at the 4- or 5-position react with overall (Z)-1,5- and 1,6-anti-stereoselectivity with respect to the hydroxy and methyl substituents.

Reactions of 5-benzyloxy-4-methylpent-2-enyl(tributyl)stannane with aldehydes promoted by bismuth(III) iodide were usefully stereoselective in favour of the (E)-1,5-anti-6-benzyloxy-5-methylalk-3-en-1-ols. Similar stereoselectivity was observed for reactions of analogous 5-benzyloxy-4-methylpent-2-enyl bromides with aldehydes when promoted by a low valency bismuth species prepared by reduction of bismuth(III) triiodide with powdered zinc so providing a “tin-free” procedure. The analogous reactions of 4-benzyloxypent-2-enyl(tributyl)stannane with aldehydes promoted by bismuth(III) iodide were also stereoselective but gave lower yields. Attempted 1,6-stereocontrol using these reactions resulted in only modest stereoselectivities. Aspects of the chemistry of the products were studied in particular their stereoselective conversion into aliphatic compounds with methyl bearing stereogenic centres at 1,5,9,13- and 1,3,5-positions along the aliphatic chain. Mechanistically, allylic organobismuth species may be involved in both sets of reactions but this was not confirmed although the similar stereoselectivities observed for both the bismuth(III) iodide mediated reactions of the pent-2-enylstannanes and the low-valency bismuth promoted reactions of the pent-2-enyl bromides are consistent with participation of similar intermediates.

The 20-deoxybryostatin 40 has been prepared using a modified Julia olefination to form the 16,17-double-bond, followed by macrolactonisation, selective deprotection and oxidation. This is the first total synthesis of a 20-deoxybryostatin.

The tin(IV) bromide promoted reaction of 7-hydroxy-7-phenylhept-2-enyl(tributyl)stannane 11 with benzaldehyde gave a mixture of the epimeric 1,8-diphenyloct-3-ene-1,8-diols 12 and so indirect methods were developed for aliphatic 1,8-stereocontrol to complete diastereoselective syntheses of (±)-patulolide C 1 and (±)-epipatulolide C 40. (5Z)-3,7-syn-7-(2-Trimethylsilylethoxy)methoxyocta-1,5-dien-3-ol 17 was prepared from the tin(IV) chloride promoted reaction of 4-(2-trimethylsilylethoxy)methoxypent-2-enyl(tributyl)stannane 16 with acrolein (1,5-syn:1,5-anti = 96:4). An Ireland–Claisen rearrangement of the corresponding benzoyloxyacetate 21 with in situ esterification of the resulting acid using trimethylsilyldiazomethane gave methyl (4E,7Z)-2,9-anti-2-benzyloxy-9-(2-trimethylsilylethoxy)methoxydeca-4,7-dienoate 22 together with 10–15% of its 2,9-syn-epimer 26, the 2,9-syn-:2,9-anti-ratio depending on the conditions used. An 88:12 mixture of esters was taken through to the tert-butyldiphenylsilyl ether 38 of (±)-patulolide C 1 together with 6% of its epimer 39, by reduction, a Wittig homologation and deprotection/macrocyclisation. Following separation of the epimeric silyl ethers, deprotection of the major epimer 38 gave (±)-patulolide C 1. The success of 2,3-Wittig rearrangements of allyl ethers prepared from (5Z)-3,7-syn-7-(2-trimethylsilylethoxy)methoxyocta-1,5-dien-3-ol 17 was dependent on the substituents on the allyl ether. Best results were obtained using the pentadienyl ether 56 and the cinnamyl ether 49 that rearranged with >90:10 stereoselectivity in favour of (1E,5E,8Z)-3,10-syn-1-phenyl-10-(2-trimethylsilylethoxy)methoxyundeca-1,5,8-trien-3-ol 50. This product was taken through to the separable silyl ethers 38 and 39, ratio 7:93 by regioselective epoxidation and alkene reduction using diimide, followed by deoxygenation, ozonolysis, a Wittig homologation and selective deprotection/macrocyclisation. Deprotection of the major epimer 39 gave (±)-epipatulolide C 40.

Reactions of the allyltin trichloride 45 generated from (4S)-4-benzyloxypent-2-enyl(tributyl)stannane 1 with imines prepared from glyoxylates proceed with useful levels of 1,5-stereocontrol in favour of (4E)-2,6-anti-2-(alkylamino)-6-benzyloxyhept-4-enoates 49. This stereoselectivity, controlled by the chirality of the stannane, dominates over any intrinsic stereochemical bias of the imine although a small amount of matching and mis-matching was observed. The allyltin trichloride 77 prepared from (4S)-4-(tert-butyldimethylsilyloxy)pent-2-enyl(tributyl)stannane 52 reacts with 1-alkoxycarbonylimines with the opposite 1,5-stereoselectivity to give the (4E)-2,6-syn-diastereoisomers 79. Matching and mismatching was more pronounced for tin(IV) chloride mediated reactions of (4R)-5-benzyloxy-4-methylpent-2-enyl(tributyl)stannane 80 with chiral 1-alkoxycarbonylimines but useful stereoselectivity in favour of (4E)-2,6-syn-2-alkyl- and arylthio-amino-7-benzyloxy-6-methylhept-4-enoates 177 was observed for reactions with achiral imines and similar, but reduced, stereoselectivity was observed for the 5-tert-butyldimethylsilyloxypentenylstannane 82. However, excellent 1,5-stereocontrol in favour of the (4E)-2,6-anti-isomers 179 was found using the 4,5-bis-alkoxypent-2-enylstannane 106. Modest (4E)-2,7-anti-stereoselectivity was observed in the analogous tin(IV) bromide mediated reactions of (S)-5-methoxy- and (S)-5-hydroxyhex-2-enyl(tributyl)stannanes (SSSSSSSSSSSS)-123 and (SSSSSSSSSSSS)-122 with achiral 1-alkoxycarbonylimines but in this series the intrinsic stereochemical bias of the imine controls the facial selectivity of reactions of chiral 1-alkoxycarbonylimines. Useful (4E)-2,6-anti-stereoselectivity was also observed in the tin(IV) chloride promoted reaction of the 4-benzyloxypent-2-enylstannane 1 with an oxime O-benzyl ether.

Two syntheses of the C(7)–C(16)-fragment 41 of epothilone D 2 were developed that were based on tin(IV) bromide mediated reactions of 5,6-difunctionalised hex-2-enylstannanes with aldehydes. In the first synthesis, (5S)-6-tert-butyldimethylsilyloxy-5-hydroxy-2-methylhex-2-enyl(tributyl)stannane 20 was reacted with (E)-but-2-enal to give (2S,7R,4Z,8E)-1-tert-butyldimethylsilyloxy-5-methyldeca-4,8-diene-2,7-diol 26 containing ca. 20% of its (7S)-epimer. Following desilylation, the crystalline (2S,7R)-triol 32 was protected as its acetonide 33 and esterified to give the (4-methoxybenzyloxy)acetate 34. An Ireland–Claisen rearrangement of this ester gave methyl (2R,3S,10S,4E,7Z)-3,7-dimethyl-10,11-(dimethylmethylene)dioxy-2-(4-methoxybenzyloxy)undeca-4,7-dienoate 35 that was converted into (2S,9S,6Z)-2,6-dimethyl-9,10-(dimethylmethylene)dioxydec-6-en-1-ol 41 by regioselective alkene manipulation, ester reduction and cleavage of the resulting terminal diol 40 with a reductive work-up. The second synthesis involved the tin(IV) bromide mediated reaction between the stannane 20 and (3S)-4-(4-methoxybenzyloxy)-3-methylbutanal 44 that gave (2S,7S,9S,4Z)-1-tert-butyldimethylsilyloxy-5,9-dimethyl-10-(4-methoxybenzyloxy)dec-4-ene-2,7-diol 45 containing ca. 20% of its (7R)-epimer. After desilylation and protection of the vicinal diol as its acetonide 46, a Barton–McCombie reductive removal of the remaining hydroxyl group gave the (2S,9S,6Z)-2,6-dimethyl-9,10-(dimethylmethylene)dioxydec-6-en-1-ol 41 after oxidative removal of the PMB-ether. The first of these syntheses uses just one chiral starting material, but the second is shorter and more convergent. It was therefore modified by the use of (5S)-6-tert-butyldimethylsilyloxy-5-(2-trimethylsilylethoxy)methoxy-2-methylhex-2-enyl(tributyl)stannane 49 that reacted with (3S)-4-(4-methoxybenzyloxy)-3-methylbutanal 44 to give a 50:50 mixture of the C(4)-epimers of (2S,9S,6Z)-10-tert-butyldimethylsilyloxy-1-(4-methoxybenzyloxy)-2,6-dimethyl-9-(2-trimethylsilylethoxy)methoxydec-6-en-4-ol 50 with high fidelity for formation of the (Z)-alkene. Following the Barton–McCombie deoxygenation, the product 52 was taken through to (2S,9S,6Z,10E)-2,6,10-trimethyl-11-(2-methyl-1,3-thiazol-4-yl)-9-(2-trimethylsilylethoxy)methoxyundeca-6,10-dienal 59 that corresponded to the fully functionalised C(7)–C(17) fragment of epothilone D 2. A precedented stereoselective aldol condensation followed by O-protection, selective deprotection, oxidation and macrocyclisation then gave the macrolide 71 that was deprotected to complete a synthesis of epothilone D 2. Finally regio- and stereo-selective epoxidation gave epothilone B 1.