Conquering peaks and illuminating depths: developing stereocontrolled organic reactions to unlock nature's macrolide treasure trove

Nelson Y. S. Lam† ^ab, Tegan P. Stockdale† ^a, Matthew J. Anketell† ^a and Ian Paterson *^a
aYusuf Hamied Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK
^bThe Scripps Research Institute, 10550 N Torrey Pines Road, La Jolla, California 92037, USA

First published on 23rd February 2021

---

---

Nelson Y. S. Lam

Nelson Lam received his BSc (Hons) degree (2015) from the University of Auckland, working in the lab of Prof. Christian Hartinger. He obtained his PhD at the University of Cambridge (2019) under the supervision of Prof. Ian Paterson as a Woolf Fisher Scholar, where he worked on the total synthesis of polyketide natural products. He is currently a postdoctoral associate and Lindemann Fellow with Prof. Jin-Quan Yu at the Scripps Research Institute, developing novel Pd-catalysed C(sp3)–H functionalisation transformations.

Tegan P. Stockdale

Tegan Stockdale received her BSc (Hons) and LLB (Hons) degrees (2016) from the University of Queensland, working in the labs of Prof. James de Voss and Prof. Joanne Blanchfield on the bioavailability of steroidal saponin natural products, isolation of bioactive products from traditional medicinal plants and synthesis of mechanistic probe molecules for cytochrome P450 enzymes. She is currently pursuing her PhD at the University of Cambridge under the supervision of Prof. Ian Paterson as a Herchel Smith Scholar, embarking on the total synthesis of the stereochemically-ambiguous natural product hemicalide.

Matthew J. Anketell

Matthew Anketell received his BA and MA from the University of Cambridge, reading the Natural Sciences Tripos. In 2015, he joined the Paterson group where he embarked on the synthesis of the stereochemically-ambiguous patellazole marine natural products. He then continued research in the same group as a Herchel Smith Scholar, where he completed the total synthesis of the actinoallolide class of anti-trypanosomal macrolides, to obtain his PhD (2020). He is currently a postdoctoral associate with Prof. Robert Britton at Simon Fraser University, Canada.

Ian Paterson

Ian Paterson's father was a master builder who shared with him the conviction that one of the best jobs you can do is a creative one, seeding his ambition to become a molecule maker. Enthralled by the aesthetic qualities of natural product structures and their biological function, his career has focused on the development of new synthetic methodology and its application in total synthesis to unlock nature's polyketide treasure trove. He received his BSc (Hons) degree from the University of St Andrews and PhD from the University of Cambridge, working with Prof. Ian Fleming. After a postdoctoral period with Prof. Gilbert Stork at Columbia University, he joined the faculty at University College London. In 1983, he returned to Cambridge, where he is Professor of Organic Chemistry and a Fellow of Jesus College. His research achievements have been recognised by various awards, and he is a Fellow of the Royal Society and the Royal Society of Edinburgh.

---

1. Introduction

The varied landscape of nature has shaped the evolution of a range of organisms large and small. Through their adaptations, which are fundamentally chemical in origin, they have developed intricate mechanisms to survive and thrive. This has resulted in the evolution of an equally diverse and varied landscape of natural products found within organisms that live from the highest mountains down to the deepest ocean trenches. Compared to nature's timescale, humanity's ability to manipulate the chemical world stands as a miniscule flash of time, differing by over six orders of magnitude! And yet in the last 200 years, the staggering advances made across the chemical sciences have produced powerful synthetic tools to target the privileged chemical space that natural products occupy, giving an unmatched opportunity to exploit their biological function in drug discovery.¹

Like artists inspired by the majesty of nature, we were drawn to the exquisite 3D form and function of these intricate natural products, with polyketides being one important bioactive class that proved particularly alluring. Belying their complex yet flexible molecular scaffolds, featuring characteristic oxygenation patterns and rich stereochemical detail, is a deceptively straightforward means for their biosynthesis engineered by polyketide synthases. This ubiquitous pathway has inspired the development of a similarly robust methodology for their total synthesis in the laboratory, generating a toolbox of stereoselective aldol reactions to replicate nature's biosynthetic machinery.

Over the last 30 years, these powerful molecular construction tools have enabled the successful completion of the total synthesis of over 40 bioactive polyketide natural products in our group. Throughout each journey, we have traversed diverse chemical terrains, with each synthetic campaign bringing its own unique set of challenges, rewards and new knowledge. Like an intrepid explorer fondly looking back on their most memorable adventures, this Feature Article recounts a personal selection of career highlights.² We chronicle the winding paths followed, hurdles overcome and lessons gleaned from the total synthesis of a select group of macrocyclic polyketides of marine and terrestrial origin: swinholide A (1994), spongistatin 1 (2001), spirastrellolide A methyl ester (2008, 2012), leiodermatolide (2014), chivosazole F (2017) and actinoallolide A (2020).

2. Asymmetric boron-mediated aldol methodology for the construction of complex polyketide natural products

To preface how we approached the total synthesis of these macrolide natural products, we must first introduce the toolbox of aldol reactions developed to enable the construction of complex polyketide structures. The biosynthesis of complex polyketides is elegantly engineered by polyketide synthases. This operates via an iterative stepwise sequence of chain extension, based on a decarboxylative Claisen-like condensation and ketone reduction, to configure the vicinal methyl-bearing and carbinol stereocentres (Scheme 1).

In a laboratory setting, it was envisaged to replicate this process through a directed asymmetric aldol reaction between a methyl or ethyl ketone and aldehyde, setting one or two stereocentres in a diastereo- and/or enantioselective manner.^3,4 The resulting β-hydroxyketone can then undergo a controlled 1,3-syn or 1,3-anti reduction to configure up to three contiguous stereocentres.⁵ These aldol-based bond constructions would then enable the controlled synthesis of diverse polyoxygenated chiral building blocks and facilitate the coupling of advanced fragments.⁶ In contrast to the enforced linearity of the biosynthetic polyketide assembly lines, the chemical synthesis would also make possible the convergent construction of complex intermediates in total synthesis.

For the boron-mediated aldol reactions of ethyl ketones, the relative configuration of the methine and carbinol stereocentres principally arises from the selectivity of the initial enolisation step, where Z-enolates produce syn-adducts while E-enolates produce anti-adducts (Scheme 2).⁶ In general, enolate geometry can be divergently controlled using a suitable combination of boron Lewis acid and tertiary amine base. Small ligands and a good leaving group on boron, combined with a bulky amine base (e.g. nBu₂BOTf, iPr₂NEt), promote the selective formation of Z-enolates. Conversely, sterically demanding ligands and a poor leaving group on boron, combined with a small amine base (e.g. cHex₂BCl, Et₃N), promote the selective formation of E-enolates. Methyl ketones generally undergo regioselective enolisation to generate the less substituted enolate. For both methyl and ethyl ketones, the resulting boron enolates react with aldehydes through a highly-ordered six-membered cyclic transition state, which is sensitive to steric and electronic influences.^4,7

In developing asymmetric boron-mediated aldol methodology for complex polyketide synthesis (Scheme 3), we addressed achieving π-facial selectivity by: (a) substrate-based control, using a chiral enolate; (b) auxiliary-based control, using an enolate with a cleavable chiral directing group; (c) reagent-based control, arising from the steric influence of chiral ligands on boron.⁷ We realised that the chiral pool Roche ester, which is commercially available in both enantiomeric forms, can be readily transformed into the corresponding methyl and ethyl ketones, enabling incorporation of a defined methyl-bearing stereocentre and reliable 1,4-syn selectivity in boron-mediated aldol reactions of type (a).⁸

Building on these findings, we next developed a set of lactate-derived ketones, which offer a complementary means of generating aldol adducts of type (b) that can be manipulated in various ways, including conversion into protected β-hydroxyaldehydes.⁹ Finally, we pioneered the application of convenient Lewis acids bearing isopinocampheyl (Ipc) ligands on boron to asymmetric aldol reactions of type (c).^7,10 This important development allowed the reinforcement or even the reversal of the intrinsic π-facial selectivity of the substrates, as well as enabling the practical synthesis of enantiomerically enriched aldol adducts from achiral ketone and aldehyde components. Collectively, these reliable and versatile construction tools afford controlled access to a diverse range of β-hydroxyketones, expediently addressing the synthetic challenges posed by macrolide and other polyketide natural products. In the remainder of this Article, the application of these reactions in total synthesis is showcased in the context of a representative selection of six bioactive macrolides completed in the group.

3. Total synthesis of bioactive macrolides enabled by aldol methodology

3.1 Swinholide A

Swinholide A (1, Scheme 4) was first reported in 1985 by Carmely and Kashman, following its isolation from the marine sponge Theonella swinhoei.¹¹ Originally misassigned as a monomeric 22-membered macrolide, subsequent X-ray crystallographic analysis by the Kobayashi/Kitagawa group revealed the C₂-symmetric, dimeric structure of this unusually large 44-membered macrodiolide and permitted the secure configurational assignment of the 30 stereocentres.¹² Swinholide A showed potent cytotoxicity across a number of cancer cell lines,¹³ with actin determined to be its cytoskeletal target.¹⁴ Interestingly, it shares a high degree of structural homology with various other macrolides of both marine and terrestrial origin, providing evidence for its production by symbiotic heterotrophic bacteria.¹⁵

Drawn by its intriguing molecular architecture and potent bioactivity, swinholide A was identified as an attractive target¹⁶ to showcase the nascent aldol methodology developed in the group.^17–24 As such, this project stands out as an early career highlight. We developed a flexible modular strategy, where the related monomeric units 2 and 3 would be constructed from fragments 4 and 5, while installation of the 44-membered macrodiolide relied on an adventurous site-selective esterification and macrocyclisation. This permitted two alternative sequences of fragment assembly to be explored to determine the optimum route. Four strategic aldol disconnections were identified as highlighted.

Preparation of C1–C15 aldehyde 4 commenced with the asymmetric construction of the dihydropyran ring (Scheme 5).¹⁷ An (+)-Ipc₂BCl-mediated aldol reaction between methyl ketone 6 and aldehyde 7 gave β-chloroenone 8, which was cyclised to provide dihydropyranone 9. Subsequent Luche reduction and Lewis acid-mediated alkylation of silyl enol ether 10 proceeded, via a Ferrier-type rearrangement, to afford aldehyde 11. Next, a vinylogous Mukaiyama aldol reaction between silyl dienol ether 12 and aldehyde 11, promoted by BF₃·OEt₂, selectively provided 1,3-anti adduct 13, as predicted by the Evans polar model.¹⁸ Subsequent manipulation, which included an HWE olefination, finally gave C1–C15 aldehyde 4.

Construction of the corresponding C19–C32 aldehyde 5 (Scheme 6)¹⁹ began with a Sharpless asymmetric epoxidation with kinetic resolution of racemic allylic alcohol 14 to give, after hydroxyl-directed reduction, 1,3-diol 15. Ozonolysis of the alkene, followed by cyclisation and methylation, gave acetal 16. This underwent a TMSOTf-catalysed allylation with allyltrimethylsilane to set the C27 stereocentre, followed by elaboration into aldehyde 17. A cHex₂BCl-mediated aldol reaction, between Roche ester-derived ethyl ketone 18 and aldehyde 17, afforded anti adduct 19. This underwent an Evans–Saksena 1,3-anti reduction and silylation to produce alkene 20. Subsequent hydroboration with thexylborane, followed by a site-specific deoxygenation sequence, then afforded the fully elaborated C19–C32 aldehyde 5, with efficient substrate-controlled installation of the five contiguous stereocentres.

Moving forward, investigation of the cHex₂BCl-mediated aldol coupling between C16–C32 ethyl ketone 21 (Scheme 7) derived from aldehyde 5 with C1–C15 aldehyde 4 revealed only a moderate level of diastereoselectivity for anti adduct 22.²⁰ Although this product could be inverted at C15 and elaborated into monomeric unit 23,²¹ this less than satisfactory result detracted from the otherwise excellent level of stereocontrol achieved. Fortunately, an alternative fragment coupling proved superior, where C1–C15 aldehyde 4 was first submitted to a Brown crotylation, followed by elaboration into methyl ketone 24 (Scheme 8). The complex aldol coupling between the silyl enol ether derivative of 24 and 5 was then best performed under Mukaiyama conditions, mediated by BF₃·OEt₂. This efficiently gave adduct 25, as predicted by the Felkin–Anh model. Finally, a 1,3-syn reduction of 25 installed the C17 stereocentre, followed by conversion into PMP acetal 23.

At this advanced stage of the campaign, an adventurous site-selective dimerisation of monomeric unit 23 was required to construct the signature macrodiolide of swinholide A. In practice, 23 was differentially desilylated under fluorous conditions to give diol 2, while ester hydrolysis gave acid 3. Initial studies revealed that site-selectivity was sensitive to the esterification conditions employed.²² Notably, Yamaguchi conditions resulted in preferential esterification at the desired C21 alcohol. Thus, diol 2 and acid 3 were first selectively coupled to give C21′ ester 26. Gratifyingly, after its controlled conversion into seco acid 27, a site-selective Yamaguchi macrolactonisation, at the desired C21 alcohol, served to assemble the desired 44-membered macrodiolide core. Finally, an uneventful global deprotection step enabled the first total synthesis of swinholide A (1) in 25 steps longest linear sequence (LLS) and 0.4% yield.^23,24

3.2 Spongistatin 1/altohyrtin A

The spongistatins (also known as the altohyrtins) are a family of extremely potent antimitotic macrolides. In 1993, members of this family were independently reported to be isolated from marine sponges by the groups of Pettit,²⁵ Kobayashi/Kitagawa²⁶ and Fusetani.²⁷ While there was initially variation in the configurational assignments, the first total synthesis by the Evans group,²⁸ followed by that of the Kishi group,²⁹ confirmed the assignment of Kobayashi/Kitagawa and established that altohyrtins A and C were identical to spongistatins 1 and 2, respectively. Structurally, spongistatin 1 (28, Scheme 9) features an elaborate 51-carbon backbone, bearing 24 stereocentres, and incorporates a highly substituted 42-membered macrolactone with a chlorotriene side chain. The macrolactone core itself encompasses the AB spiroacetal, the CD spiroacetal and the bis-tetrahydropyran EF rings.

This macrolide chemotype is amongst the most potent compounds to be tested in the NCI panel of human carcinoma cell lines, with exemplary picomolar cytotoxicity. Furthermore, in vivo human carcinoma xenograft studies showed curative responses for ovarian tumours and melanoma at extremely low dosage levels.³⁰ Recent X-ray crystallographic studies of the spongistatin 1/tubulin complex revealed that the spongistatins bind to the maytansine domain of tubulin,³¹ which prevents microtubule assembly, thereby inhibiting mitosis.³² Despite such promising preclinical results, their meagre and unsustainable natural supply stalled further development. Augmented by its impressive bioactivity and the extreme paucity in natural supply, the extraordinary molecular architecture of spongistatin 1 renders it a compelling synthetic target.³³ Falling under its spell, we viewed it as the definitive complex macrolide to showcase the aldol methodology developed in the group.³⁴ We aimed to devise a flexible and convergent synthesis that was efficient enough to deliver useful quantities and permit further biological studies.

From the outset of this ambitious campaign, the key concern was how to achieve both high levels of stereocontrol and chemoselectivity, in managing the array of potentially reactive functionalities. In particular, the correct choice of protecting groups and ordering of functional group transformations would be vital in engineering the construction and coupling of the highly oxygenated intermediates. After several iterations, our strategy evolved to target three key fragments – C1–C15 AB spiroacetal 29, C16–C28 CD spiroacetal 30 and C29–C51 EF bis-tetrahydropyran 31. This plan entailed a late-stage macrolactonisation, a Wittig reaction to install the C28–C29 olefin, and an aldol coupling reaction to forge the C15–C16 bond and associated stereocentres. We envisaged that an expedient synthesis could be enabled through a carefully choreographed sequence of nine boron-mediated aldol reactions. Consequently, spongistatin 1 stands out as a second career highlight.

Assembly of C1–C15 AB spiroacetal 29 (Scheme 10) included three boron-mediated aldol reactions, in which reagent control with (−)-Ipc₂BCl was used to reinforce any substrate-based stereoinduction.³⁵ To begin, a modified Brown allylation of aldehyde 32, mediated by (+)-2-carene derived ligands, gave homoallylic alcohol 33. After conversion into aldehyde 34, an aldol reaction with acetone, followed by silylation, generated ketone 35. Synthesis of the corresponding C9–C15 partner 36 began with an aldol reaction between ketone 37 and aldehyde 38 to give 1,4-syn adduct 39. Although this led to the epimeric configuration at C11, the excellent yield and selectivity rendered a downstream invertive process the preferred means to obtain the desired C11 stereocentre. This approach entailed protecting group manipulation and a Takai methylenation, before a Mitsunobu reaction gave inverted alcohol 40, which was converted into aldehyde 36.

A complex boron-mediated aldol coupling reaction was now conducted between chiral methyl ketone 35 and chiral aldehyde 36. Employing (−)-Ipc₂BCl/Et₃N, this gave exclusively 1,5-anti adduct 41. The excellent stereocontrol is attributed to the synergistic combination of the 1,3-syn preference of 36, the 1,5-anti preference³⁶ of 35 and the influence of the chiral reagent.⁷ Next, selective desilylation of ketone 41 served to construct the doubly anomerically-stabilised, axial–axial, AB spiroacetal 42. Oxidation, followed by an axial Grignard addition, then gave alcohol 43. Finally, a sequence of fine-tuning of protecting groups and oxidation state adjustment afforded C1–C15 aldehyde 29.

The synthesis of the corresponding C16–C28 fragment 30 proved to be a hurdle, owing to the singly anomerically-stabilised, axial–equatorial, CD spiroacetal.³⁷ An initially explored kinetic approach proved disappointing, as only a slight preference for the desired spiroacetal configuration was observed. Ultimately, an equilibration approach (Scheme 11) became the preferred route, with the stereocentres introduced by three reagent-controlled allylations and an aldol reaction. This commenced with a Brown allylation of aldehyde 44 to give alcohol 45. Following methylation and ozonolysis, a further allylation of aldehyde 46 gave alcohol 47. This was then elaborated into methyl ketone 48, in readiness for a downstream aldol coupling. Synthesis of its aldehyde partner 49 employed a modified Brown allylation, from aldehyde 50, to generate homoallylic alcohol 51. This was followed by silylation and oxidative alkene cleavage to give aldehyde 49.

A complex aldol coupling between chiral methyl ketone 48 and chiral aldehyde 49 was then performed using (−)-Ipc₂BCl/Et₃N, proceeding with triple asymmetric induction to provide solely adduct 52. From here, desilylation and concomitant spiroacetalisation, followed by acetal equilibration, under acidic conditions, gave an equimolar mixture of 53 to 54. At this juncture, chromatographic separation, followed by re-equilibration of the undesired spiroacetal 54, allowed gram quantities of CD spiroacetal 53 to be accumulated. For the remainder of the campaign, there was a risk that unanticipated re-equilibration of the C23 configuration might occur. Hence, any acidic conditions needed to be avoided. Following conversion into aldehyde 55, a Grignard addition and oxidation then produced the complete C16–C28 ketone 30.

The synthesis of the C29–C51 EF bis-tetrahydropyran motif of spongistatin constituted a logistical challenge. This was both due to the range of potentially reactive functionalities present and the increased stereochemical complexity relative to the other fragments.^34,38 Given that the planned coupling between the northern and southern hemispheres involved a Wittig reaction with an advanced C1–C28 aldehyde, an efficient construction of phosphonium salt 31 was sought (Scheme 12). The boron-mediated aldol reactions used to assemble this fragment enabled the expedient formation of four key carbon–carbon bonds and six associated stereocentres. Thus, synthesis of the E ring precursor began with a cHex₂BCl-mediated aldol reaction between lactate-derived ketone 56 and 5-chloropentanal to give syn adduct 57. This was then elaborated into aldehyde 58, in anticipation of a further aldol coupling reaction.

Synthesis of F ring coupling partner 59 commenced with a substrate-controlled cHex₂BCl-mediated aldol reaction between Roche ester-derived ketone 60 and acetaldehyde to give anti adduct 61. An Evans–Saksena reduction then gave the corresponding 1,3-anti diol, which was transformed into aldehyde 62. Following HWE olefination to give enoate 63, a reagent-controlled dihydroxylation under Sharpless conditions³⁹ delivered diol 64. Further protecting group and oxidation state manipulation gave aldehyde 65, enabling a second HWE olefination to give enone 66.⁴⁰ At this point, acid-mediated acetonide cleavage induced a hetero-Michael addition to generate a mixture of tetrahydropyrans. These were equilibrated under basic conditions, to produce the required all-equatorial F ring 67.

Once the chlorotriene moiety was installed, it was desirable to minimise the number of transformations undertaken in the presence of this delicate moiety. Thus, elaboration at C36 of 59 and coupling with aldehyde 58, followed by introduction of the full side chain, was planned. To this end, the C45 ketone was transiently masked through methylenation, followed by oxidation at C37 to afford methyl ketone 59. The aldol construction of the C35–C36 bond proved to be challenging, attributed to steric congestion inhibiting the enolisation of 59. This necessitated the use of cHex₂BBr as a more reactive Lewis acid. After coupling with aldehyde 58, this afforded adduct 68 as predicted by the Felkin–Anh model. Subsequent silylation and acid-catalysed cyclisation then produced E-ring acetal 69, which was elaborated into ketone 70. Next, installation of the chlorotriene moiety in 71 was secured by a substrate-controlled cHex₂BCl-mediated aldol reaction, between ketone 70 and aldehyde 70a. Unexpectedly, the 1,5-anti stereoinduction³⁶ normally seen in related aldol reactions of simpler β-alkoxy methyl ketones was reversed, ascribed to the opposing influence of the proximate F ring motif.⁴¹ Fortunately, this serendipitous outcome instead gave the required 1,5-syn adduct 71, which was converted into phosphonium salt 31. Notably, the primary chloride survived throughout, prior to its substitution with triphenylphosphine, without encountering any chemoselectivity issues.

With all three major fragments secured, their controlled coupling and the pivotal macrolactonisation step were pursued.^34,41 The substrate-controlled fragment coupling of C1–C15 aldehyde 29 and C16–C28 ethyl ketone 30 is one of the most complex examples of a boron-mediated aldol reaction (Scheme 13).⁴² Use of cHex₂BCl/Et₃N to generate the E-enolate from 30 and addition to 29 proceeded smoothly to give anti adduct 72, installing the C15 and C16 stereocentres with high fidelity. Following conversion into aldehyde 73, the challenging Wittig reaction, with phosphonium salt 31, proved to be an additional hurdle to overcome. After extensive experimentation, a rigorously de-oxygenated solvent mixture of THF/HMPA was found to be optimum, along with treatment of phosphonium salt 31 with CaH₂, prior to deprotonation (LiHMDS) and addition of aldehyde 73. This protocol enabled the Wittig coupling to reproducibly proceed in good yield, to give solely Z-alkene 74. From here, a sequence of PMB ether and TCE ester cleavage produced seco acid 75. Under Yamaguchi conditions, 75 reacted site-selectively, at the C41 alcohol, to give exclusively 42-membered macrolactone 76. Finally, a global deprotection concluded this eventful synthetic saga and delivered spongistatin 1 (28) in 33 steps LLS and 1.6% yield.

The resulting supply of synthesised spongistatin 1 augmented the meagre quantity isolated from >400 kg of sponge by the Pettit group, enabling the continuation of its biological evaluation, including preliminary SAR studies and mapping of its tubulin binding site.³¹ Serendipitously, a side-product with a dehydrated E ring, obtained in the final deprotection step, transpired to be an even more potent cytotoxic agent than the parent natural product.⁴³

3.3 Spirastrellolide A methyl ester

Spirastrellolide A was first reported in 2003 by the Andersen group, following its isolation from the Caribbean sponge Spirastrella coccinea.⁴⁴ Characterised as its methyl ester 77 (Scheme 14), it showed potent antiproliferative activity against cancer cell lines and acted as a selective inhibitor of Ser/Thr phosphatase 2A.⁴⁵ Structurally, it features a 47-carbon skeleton and 21 stereocentres, with a 38-membered macrolide, bearing a skipped diene side chain. The macrocyclic core contains a tetrahydropyran (A ring), a 6,6-spiroacetal (BC rings) and a chlorinated 5,6,6-bis-spiroacetal (DEF rings).

The promising biological properties of spirastrellolide A and its natural scarcity, along with a complex molecular architecture reminiscent of spongistatin and an incomplete stereochemical assignment, provided compelling reasons for pursuing its total synthesis.^46–48 As such, spirastrellolide stands out as a third career highlight. These studies evolved alongside the ongoing structural elucidation of this intriguing marine macrolide by the Andersen group.

The stereochemical ambiguities demanded a flexible modular approach, whereby each of the three fragments 78, 79 and 80 were to be constructed with versatile coupling handles. Installation of the side chain, by attachment of stannane 81 to a preformed macrocycle derived from seco acid 82, was planned. Two key disconnections were made: at C16–C17 to disassemble the BC spiroacetal, to reveal fragments 80 and 83; and at C24–C25 in 84 to reveal fragments 78 and 79.⁴⁹ Four aldol bond constructions were identified for access to these fragments.

Synthesis of C1–C16 alkyne 80 (Scheme 15)⁵⁰ commenced with a Grignard addition into epoxide 85, followed by olefin cross-metathesis with methyl acrylate, to give enoate 86. This enabled a hetero-Michael addition to install the cis-tetrahydropyran A ring in 87. After conversion into the corresponding C9 aldehyde, a chelation-controlled Hosomi–Sakurai allylation, mediated by TiCl₄, gave alcohol 88. The derived C11 aldehyde 89 then underwent a reagent-controlled aldol reaction, mediated by (−)-Ipc₂BCl, with methyl ketone 90 to afford 1,4-syn adduct 91. Following an Evans–Saksena reduction, this was manipulated to give vinyl dibromide 92. Finally, treatment with nBuLi allowed controlled conversion into alkyne 80, affording a robust route adaptable to a gram-scale synthesis. The route to C17–C24 vinyl iodide 78 commenced with hydrotitanation/iodination of alkyne 93 (Scheme 16).⁵¹ The derived aldehyde 94 then underwent an Evans glycolate aldol addition with 95 to produce syn adduct 96. Transamination, followed by silylation and allylation, gave ketone 97. A chelation-controlled ketone reduction, mediated by Zn(BH₄)₂, set the final C20 stereocentre. Further adjustment then delivered C17–C24 vinyl iodide 78.

The synthesis of the DEF-containing bis-spiroacetal fragment 79 (Scheme 17) underwent continuous refinement during the course of the campaign.^52–54 Recognising that the bis-spiroacetal is doubly stabilised by the anomeric effect suggested cyclisation from a suitable linear precursor under acidic conditions. To streamline the route, the installation of both sets of diols in a single step was proposed, using a Sharpless asymmetric dihydroxylation³⁹ on diene 98. Ultimately, this led to the pursuit of an adventurous strategy, centred on the implementation of a double dihydroxylation/spiroacetalisation cascade.⁵⁵

Synthesis of fragment 79 commenced via an Oehlschlager–Brown chloroallylation⁵⁶ of aldehyde 99, followed by acetal cleavage and methyl ether formation, to give syn adduct 100. The corresponding C33–C40 aldehyde 101 was prepared from aldehyde 102, utilising an Knoevenagel-type condensation with malonic acid. A cHex₂BCl-mediated aldol reaction between aldehyde 103 and ethyl ketone (S)-104 then delivered anti adduct 105, which was taken on to C33 aldehyde 101. At this point, fragment union, through an aldol coupling of methyl ketone 100 and aldehyde 101, gave C26–C40 adduct 106. This was elaborated into the required diene 98 for the projected double dihydroxylation/spiroacetalisation cascade. This reaction initially produced a complex mixture of hemiacetals 107. Gratifyingly, when exposed to mild acidic conditions, cyclisation occurred to deliver the desired DEF-containing bis-spiroacetal, which was then silylated to give 79. Notably, this expedient cascade route proved to be readily scalable, delivering gram quantities of 79.

The planned fragment union was now executed via a Suzuki coupling (Scheme 18). From 79, conversion into alkene 108 enabled hydroboration and Pd-catalysed sp²–sp³ cross-coupling with vinyl iodide 78 to give diene 84. Next, the crucial double hydroboration using BH₃·SMe₂ served to install the primary C17 alcohol, as well as correctly configuring the C23 and C24 stereocentres. This was followed by its advancement to C17–C40 aldehyde 83.

With the C17–C40 aldehyde secured, coupling with the C1–C16 fragment 80a and spiroacetalisation to complete the cyclic ether skeleton was addressed (Scheme 19). Lithiation of alkyne 80a and addition to aldehyde 83 proceeded smoothly to deliver C1–C40 alcohol 109. A Lindlar reduction and oxidation gave Z-enone 110.

On exposure to DDQ, this underwent global PMB ether cleavage at C1, C13 and C21, enabling facile cyclisation to afford spiroacetal 111. These mildly acidic reaction conditions also caused unexpected desilylation at C23 to reveal the highlighted secondary alcohol. Considered an annoyance and inconsequential at the time, this unintended deprotection turned out to be profoundly important. From 111, oxidation, followed by selective C37 desilylation, provided seco acid 112. Under Yamaguchi conditions, this delivered 38-membered macrolactone 113, in essentially quantitative yield. In our experience, this welcome result stands as a record yield for a macrolactonisation performed on a complex seco acid. The ease and striking efficiency of this crucial macrocyclisation reaction is attributed to a favourable conformational pre-organisation in 112.

From macrocycle 113, side chain attachment appeared to be the final hurdle to surmount. However, there proved to be an unanticipated sting in the tail! Frustratingly, selective desilylation at C40 resisted all attempts, necessitating a switch in protecting groups. Nevertheless, this led to a lucky break. Global desilylation of 113 afforded highly crystalline pentaol 114, which was submitted to single-crystal X-ray crystallographic analysis (Scheme 20). Gratifyingly, this confirmed the configuration of all the stereocentres that had been installed, as well as revealing the network of H-bond interactions within the macrocyclic core.⁵⁷ From here, the diols were converted into the corresponding acetonides, permitting controlled manipulation at C40 for side chain installation. Disappointingly, a number of approaches proved to be unsuccessful, which was attributable to the steric encumbrance of the macrocycle. Nonetheless, it was found that Wittig methylenation, followed by olefin cross-metathesis (Grubbs II) with alkene 116, gave allylic carbonate 117. This enabled a π-allyl Stille coupling with stannane 81 to deliver the full C1–C47 skeleton in 118. Finally, a global deprotection completed the first total synthesis of spirastrellolide A methyl ester (77),⁴⁶ and unambiguously confirmed the full 3D molecular structure for what started out as a dynamic target.

Following on from this success, we pursued a second-generation synthesis, with a focus on constructing a now fully defined target by a more direct route. We aimed to remove redundancies, such as protecting group and oxidation level manipulations, and reduce the overall step count.⁴⁷ A major economy was that this was to be pursued by a single student. In addition, we now planned to complete the entire carbon skeleton in the seco acid prior to macrolactonisation, circumventing the difficulties experienced with late-stage side chain attachment to the preformed macrocycle. The key fragments from the first-generation synthesis, vinyl iodide 78, DEF rings 79 and alkyne 80 could again be used in this more streamlined plan.

Formation of C17–C40 aldehyde 119 (Scheme 21) commenced with selective C26 desilylation of 79, oxidation and Wittig methylenation to give alkene 120. Hydroboration of 120, Suzuki coupling with vinyl iodide 78 and double hydroboration proceeded to give C17–C40 fragment 121. At this point, selective cleavage of the C17 and C37 TES ethers, and debenzylation gave the corresponding triol. As a finale, a one-pot triple oxidation at C17 and C40 afforded the revised C17–C40 fragment 119, terminating in a γ-lactone and aldehyde as coupling handles. With all the requisite fragments in hand, the completion of the fully elaborated seco acid skeleton was addressed. Frustratingly, initial attempts at reproducing the lithium acetylide addition of alkyne 80 with revised aldehyde 119 failed, due to chemoselectivity issues. Fortunately, this pivotal fragment coupling was efficiently achieved under Nozaki–Hiyama–Kishi conditions with iodoalkyne 122 and aldehyde 119 to form alcohol 123 (Scheme 22). From here, the Lindlar reduction, oxidation and PMB ether cleavage/spiroacetalisation sequence was conducted without incident. This delivered C1–C40 fragment 124, now retaining the C23 TES ether that had been unexpectedly cleaved in the first-generation synthesis.

At this juncture, the entire spirastrellolide A skeleton now seemed within our grasp. In the event, C1 desilylation, reduction of the γ-lactone and vinyl Grignard addition gave triol 125. Treatment of 125 with triphosgene, followed by oxidation, then gave acid 126, bearing a cyclic carbonate motif. Gratifyingly, a π-allyl Stille coupling between acid 126 and stannane 127 smoothly delivered the complete C1–C47 seco acid 128, with concomitant unmasking of the C37 alcohol.

To our dismay, subjecting seco acid 128 to Yamaguchi and various other macrolactonisation protocols now failed to produce any desired product! This nightmare scenario prompted a critical examination of the structural differences between the first- and second-generation substrates. After discounting the nature of the side chain being an inhibitory factor, the presence of the C23 TES ether in 128 seemed to be the likely culprit. To probe if this was responsible for this disparate reactivity, seco acid 128 was subjected to controlled desilylation to obtain either the C23 alcohol 129 or the C22–C23 diol 130. Sure enough, submitting 129 and 130 to Yamaguchi conditions led to the immediate return of reactivity, delivering macrolactones 131 and 132, respectively. With this unforeseen hurdle surmounted and the macrocycle secured, the finishing line was now in sight. From 132, controlled deprotection delivered spirastrellolide A methyl ester (77) in 23 steps LLS and 6% overall yield, representing a marked improvement on the first-generation route.⁴⁷

This demanding project highlights the unseen hand of serendipity in total synthesis endeavours. In hindsight, the successful first-generation route was enabled through the unexpected release of the C23 alcohol during the BC-spiroacetal formation. This unforeseen outcome acts as a reminder of the subtle conformational influence of even distal protecting groups. In this case, acting to mutate a pre-organised conformation favouring macrocyclisation to one that completely inhibits it.

3.4 Leiodermatolide

In 2008, the Wright group disclosed the isolation of leiodermatolide (133, Scheme 23) from the lithistid sponge Leiodermatium sp., collected off the coast of Florida.⁵⁸ This novel macrolide chemotype demonstrated potent antiproliferative and tubulin-binding activity, with a mechanism hypothesised to be orthogonal to existing tubulin-binders. From the outset, we were intimately involved with its stereochemical elucidation in collaboration with the Wright group, where detailed NMR analysis and computational NMR predictions allowed the determination of the relative configuration of the C1–C15 macrocycle and the C21–C25 δ-lactone.⁵⁹ However, the distal nature of these two stereoclusters precluded a confident assignment of the complete 3D structure, necessitating additional detective work and a focused synthetic campaign. The close involvement from structural characterisation to synthesis places leiodermatolide as a fourth career highlight.

In preliminary efforts towards assembling the 16-membered macrolactone core, a specific rotation opposite in sign to the natural product was obtained, tentatively suggesting that we were pursuing the wrong enantiomer.^60,61 Subsequently, the synthesis of two candidate diastereomers of leiodermatolide, reported by Fürstner,⁶² established the absolute configuration and the relationship between the two stereoclusters as indicated in 133. This prompted a revised strategy, disassembling leiodermatolide into two fragments 134 and 135, with a late-stage macrolactonisation.⁶³ The latter fragment could then be disassembled via a Heck reaction to construct the E,E-diene, revealing iodide 136 and δ-lactone 137. Four key aldol disconnections were identified to install the oxygenation and stereochemistry.

Synthesis of stannane 134 (Scheme 24) commenced with a Grignard addition into Weinreb amide 138, where treatment of the resulting ketone with Comins’ reagent, followed by Suzuki cross-coupling of the resulting enol triflate 139, gave E-trisubstituted alkene 140. After revealing aldehyde 141, a cHex₂BCl-mediated aldol reaction, with ethyl ketone (R)-104, configured the 7,8-anti stereocentres in adduct 142. This was transformed into alkynone 143, where iodide addition and stereoselective protonation of the intermediate allenol, followed by an Evans–Saksena reduction, configured 1,3-diol 144. From here, acetonide formation and iodine/tin exchange gave stannane 134.

Synthesis of δ-lactone 137 (Scheme 25) commenced with a cHex₂BCl-mediated aldol reaction, between lactate-derived ketone (R)-104 and propionaldehyde, to give 145. This was manipulated to afford ketone 146. A sequence of a Mukaiyama aldol reaction, lactonisation and silylation then gave fragment 137. In a similar manner, synthesis of iodide 136 commenced with a lactate aldol reaction, between ketone (S)-104 and aldehyde 147 to give adduct 148. This was converted into diol 136, which engaged with alkene 137 in a Heck reaction to generate diene 149. Following transformation into vinyl iodide 135, a pivotal Stille cross-coupling with stannane 134 afforded the full C1–C25 carbon skeleton 150 of leiodermatolide.

At this point, the derived seco acid 151 was smoothly macrolactonised, under Yamaguchi conditions, and desilylated to give triol 152, in readiness for the endgame. Frustratingly, attempted site-selective carbamoylation of triol 152 favoured the C7 position, contradicting the results from model studies and with other electrophilic reagents, which proved to be an unforeseen roadblock. After extensive experimentation, a carefully choreographed sequence emerged. This involved a transient bis-silylation of C7 and C9, selective desilylation at C9 under acidic conditions, and carbamoylation, followed by global desilylation. Gratifyingly, this afforded (−)-leiodermatolide (133) in 23 steps LLS and 3.2% yield, which proved to be identical to an authentic sample in all respects.⁵⁹

3.5 Chivosazole F

The chivosazoles are a family of actin-binding terrestrial macrolides, isolated in 1997 by Reichenbach, Höfle and co-workers from Sorangium cellulosum myxobacteria.⁶⁴ Their full 3D molecular architecture was subsequently determined by Kalesse,^65,66 based on a combination of chemical synthesis, conformational analysis and genetic analysis. This revealed chivosazole F (153, Scheme 26) to be a 31-membered macrolactone, containing an oxazole and three distinct polyene regions, along with 10 stereocentres.

At the outset, the introduction of the stereodefined polyenes with alternating geometry, which were known to readily isomerise, was identified as a key challenge. This challenge did indeed transpire and caused nightmares throughout our synthesis campaign.⁶⁷ The frustration associated with this shapeshifting molecule made our eventual success all the sweeter; placing chivosazole F as a fifth career highlight. In an initial approach, three aldol disconnections were proposed in combination with a series of Stille couplings and HWE olefination, to set the geometry of the delicate polyene regions, leading back to four fragments 154–157.

Synthesis of the C6–C13 fragment 154 began from Evans imide 158 (Scheme 27). While it had been initially planned to introduce the required 12Z-alkene via a vinylogous aldol reaction with Z-bromoacrolein, this was found to readily isomerise giving an early preview of the unanticipated problems ahead.⁶⁸ To circumvent this difficulty, a Kobayashi-type, vinylogous Mukaiyama aldol reaction,⁶⁹ with aldehyde 159, mediated by TiCl₄, afforded the anti adduct 160, which was converted into TES ether 161. The required 12Z-alkene was now installed via a palladium-mediated trans-debromination⁷⁰ to afford bromide 162. Following conversion into 163, installation of the 6Z-vinyl iodide was achieved by a Stork–Zhao olefination⁷¹ to afford C6–C13 fragment 154. Moving on to the synthesis of C14–C26 fragment 155, this commenced with an (−)-Ipc₂BCl-mediated aldol reaction between methyl ketone 164 and bromodienal 165 to afford adduct 166. Further manipulation gave acid 167, which was first coupled with amine 168, before cyclisation to form oxazoline 155. While the initial plan⁷² envisaged conversion into the oxazole at this stage, the oxidation conditions were found to be incompatible with the vinyl iodide, necessitating the postponement of this transformation to after fragment union. Synthesis of the final C27–C35 fragment 156 commenced with a cHex₂BCl-mediated aldol reaction, between ketone 169 and aldehyde 170, to provide anti adduct 171. Following transformation into the corresponding aldehyde 172, a Stork–Zhao olefination installed the required 27Z-alkene in 173 and further elaboration gave fragment 156.

For assembling the three alkene fragments 154–156 and acrylate derivative 157, a suitably choreographed sequence of Stille coupling reactions was sought. This was best accomplished via a remarkable one-pot process, involving sequential addition of 154, 155, 156 and 157 to a solution of palladium catalyst to progressively assemble 174. This exquisitely controlled transformation is possible due to the increased reactivity of the E-vinyl stannane in 157 over the Z-vinyl stannane in 154, and the increased reactivity of the iodide over the bromide in linchpin 155. At this advanced stage, we found that the C4–C9 triene motif in 174 was highly susceptible to isomerisation, precluding formation of the chivosazole macrocycle and necessitating a re-evaluation of the strategy. This frustrating roadblock led to an alternative assembly approach (Scheme 26, vide supra) based on the four fragments 155, 157, 163 and 175, and changing the planned order of coupling to form the isomerisation-prone southern tetraene as late as possible.

Synthesis of the revised C27–C35 fragment 175 (Scheme 28), incorporating a Still-Gennari-type phosphonate,⁷³ was achieved from intermediate 171, setting the stage for exploration of the revised endgame. As before, fragments 155, 163 and 175 were coupled under mild Stille conditions, either sequentially or in a one-pot operation, providing advanced phosphonate 176. This then underwent HWE olefination with aldehyde 157, followed by treatment with MnO₂ to oxidise at C7 and aromatise the oxazoline to the required oxazole in 177. From here, the long sought after 31-membered macrolactone was formed by a Stork–Zhao olefination, followed by an intramolecular Stille coupling reaction. Global desilylation then afforded chivosazole F (153) in 20 steps LLS and 2.5% yield, which to our great relief retained all of the carefully crafted alkene geometry. This unexpectedly challenging synthesis nicely showcases the power of Stille reactions to couple multiple fragments in a single pot under remarkably mild conditions to form chemically and thermally sensitive polyene systems.

3.6 Actinoallolide A

The actinoallolides are a family of terrestrial macrolides isolated in 2015 by the Inahashi group from cultured Actinoallomurus fulvus bacteria, obtained from a soil sample collected in Thailand.⁷⁴ In testing against a range of organisms, actinoallolide A (178, Scheme 29) showed highly selective, nanomolar potency against Trypanosoma parasites, with no inhibitory activity against MRC-5 human cells, highlighting it as a potential drug lead for the treatment, inter alia, of Chagas disease and African sleeping sickness. Structurally, it features a 12-membered macrolactone incorporating a five-membered hemiacetal, two trisubstituted alkenes and 10 stereocentres.

Captivated by its intriguing structure and potential as a drug lead in the treatment of neglected tropical diseases, actinoallolide A was viewed as an important target for realising an efficient total synthesis and exploring its mechanism of action. It was proposed to initially disassemble the lactone linkage and macrocyclic alkene in 178 to reveal C1–C8 alkene 179 and C9–C21 alkene 180.⁷⁵ We envisaged an adventurous ring-closing olefin metathesis (RCM) to install the highlighted trisubstituted E-alkene. However, this “do or die” key step was perceived as a high-risk, high-reward manoeuvre due to the lack of examples in constructing comparable medium-ring systems. The thrill of pursuing such an adventurous RCM approach stands as a sixth career highlight. Fragment 179 was planned to be constructed from dioxolanone 181⁷⁶ and a substrate-controlled aldol reaction. Adapting a strategy from the earlier synthesis of the ebelactones,⁷⁷ an Ireland–Claisen rearrangement was selected to configure the distal C14 stereocentre and trisubstituted E-alkene in 182. The C11–C13 region would be introduced, in turn, by a lactate aldol reaction⁹ and allylation of aldehyde 183. Lastly, the C18–C20 stereotriad in 179 would be set by a third aldol reaction.

Following this blueprint, the synthesis of side chain fragment 180 (Scheme 30) commenced with a substrate-controlled titanium-mediated aldol reaction,^78,79 between ketone (R)-169 and methacrolein, and in situ reduction to provide diol 184. Esterification afforded 185, which underwent an Ireland–Claisen rearrangement to efficiently install the desired 1,5-syn relationship, between C14 and C18, in E-alkene 182. Reduction of 182 then gave aldehyde 186, in readiness for a cHex₂BCl-mediated aldol reaction with ketone (S)-104. Here π-facial control from the enolate component overrides any inherent bias of the aldehyde to form the desired anti adduct 187. After conversion into aldehyde 183, Lewis acid-mediated allylation under Hosomi–Sakurai conditions, exploiting the Felkin–Anh model reinforced by the Evans polar model, served to complete the efficient assembly of C9–C21 fragment 180.

Synthesis of the second fragment 179 (Scheme 30) started out from dioxolanone 181, derived from (S)-lactate,⁸⁰ which underwent enolate alkylation to install the C6 stereocentre in 188. This was manipulated to provide propyl ketone 189, which underwent a substrate-controlled, lithium-mediated aldol reaction with the DMB ether-containing aldehyde 190, to afford syn adduct 191. This was followed by PMBM (4-methoxybenzyloxymethyl) ether formation under mild conditions. This somewhat unusual choice of benzylic protecting group reconciled the earlier failure of a more conventional route, based on a PMB ether transposition. After selective DMB ether cleavage to give 192, oxidation then completed the construction of acid 179.

Ahead of embarking on the risky RCM manoeuvre, fragment union (Scheme 31) between 179 and 180 was accomplished by Yamaguchi esterification. In the absence of any plan B, it was an immense relief that extended treatment (over seven days!) with Hoveyda–Grubbs II catalyst in refluxing toluene was found to cleanly promote cyclisation to provide the required E-alkene in the 12-membered macrolactone 193. Notably, this remarkable RCM reaction constitutes the most complex example to form a trisubstituted alkene in a medium-sized ring.⁸¹ In the endgame, this key macrocycle 193 was then elaborated into triketone 194, which underwent controlled desilylation and hemiacetalisation to complete the first total synthesis of (+)-actinoallolide A (178) in 20 steps LLS and 8% yield. The other four members of the actinoallolide family were readily prepared by late-stage diversification from 178. Leveraging this expedient endgame, a photoaffinity probe was also designed and synthesised,⁸² with a view to identifying the actinoallolide protein-binding target and mechanism of action.

4. Conclusions and perspectives

Spanning more than three decades of research, our group has achieved the total synthesis of over 40 distinct families of bioactive marine and terrestrial polyketide natural products. Throughout this challenging yet rewarding enterprise, we have demonstrated that chemists can emulate the sophisticated stereochemical control and modularity of the polyketide synthase biosynthetic machinery. In this context, the boron-mediated aldol reactions of chiral ethyl and methyl ketones with aldehydes are showcased in efficiently constructing a selection of densely oxygenated macrocyclic systems, ranging in ring size from 12- to 44-membered. Furthermore, the stereoinduction can be fine-tuned using the directing influence of Ipc ligands on the intermediate enolate. The reliability of these methods is further demonstrated through their varied applications in the stereocontrolled construction and coupling of large and small fragments. In conjunction with the controlled 1,3-syn or 1,3-anti reduction of the resulting β-hydroxyketones, this powerful aldol platform enables the rapid generation of stereochemical complexity and oxygenation to help unlock nature's polyketide treasure trove.

The synthetic methodology inspired by these exquisite polyketide architectures is exemplified by its application to the six macrolides covered in this article, and was first validated by the pioneering total synthesis of swinholide A. This demonstrated that carefully choreographed aldol reactions of ketones can be an effective tool to not only forge carbon–carbon bonds, but at the same time install stereocentres in a highly selective manner. Notably, the total synthesis of spongistatin 1 constitutes one of the most complex demonstrations of the utility, high efficiency and excellent levels of stereocontrol afforded by boron-mediated aldol methodology. Looking back on the serendipitous journey towards spirastrellolide, we were incredibly lucky in accidentally removing what transpired to be a problematic TES ether before the macrolactonisation step, which only became apparent in the second-generation endgame. The total synthesis of leiodermatolide relied on flexible planning, combined with a series of highly stereocontrolled aldol reactions, efficient fragment couplings and careful tactical endgame manoeuvres for eventual success. The chivosazole campaign provided a painful reminder that complex polyene regions need to be viewed with caution when planning a synthesis. Once installed, sp³-stereocentres can generally be depended on to retain their configuration. In contrast, the potential shapeshifting behaviour of conjugated alkenes can be much less predictable. As a bookend project, the actinoallolide synthesis was achieved by a single intrepid student, accomplishing the “do or die” RCM manoeuvre through sheer determination and perseverance.

On a more fundamental level, the challenging pursuit of these macrolide targets revealed so much more of the inherent chemical nature of these intriguing compounds than could ever be conjectured by pen and paper or thought experiments. Moreover, the new knowledge gained from embarking on their total synthesis reinforces the symbiotic relationship between synthetic and natural product isolation chemists, further enriching our collective understanding of these intriguing compounds.

Though not highlighted in this article, the group's versatile aldol methodology has also enabled the progression of polyketide drug candidates into clinical trials, as exemplified by the landmark Novartis large-scale total synthesis of discodermolide.⁸³ These trials, tribulations and serendipitous discoveries underscore how much we still have to learn to engineer precise molecular manipulations, especially when faced with such a dazzling array of functionalities that exemplify these and other natural product targets. Despite all this, the remarkable total syntheses of swinholide A, spongistatin 1, spirastrellolide A methyl ester, leiodermatolide, chivosazole F and actinoallolide A demonstrate how far we have progressed.

In closing, it is a pleasure to warmly thank the natural product groups that have discovered these fascinating macrolides, keeping us gainfully employed and intellectually challenged throughout. Without their timely bioprospecting activities and structural elucidation work, we would not have been able to embark on these and other equally memorable synthetic adventures.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The work summarised in this Feature Article encompasses over 30 years of research, enabled by a total of 94 highly talented doctoral students and 60 postdocs at Cambridge. Their dedication, insight, perseverance and creative contributions are gratefully acknowledged. We thank the Woolf Fisher Trust (scholarship to NYSL) and the Herchel Smith Fund (scholarship to TPS and MJA) for support.

Notes and references

1. I. Paterson and E. A. Anderson, Science, 2005, 310, 451–453 CrossRef PubMed .
2. For the first part of this personal retrospective, see: T. P. Stockdale, N. Y. S. Lam, M. J. Anketell and I. Paterson, Bull. Chem. Soc. Jpn., 2021, 94, 713–731 CrossRef CAS .
3. I. Paterson, Pure Appl. Chem., 1992, 64, 1821–1830 CAS .
4. A. Bernardi, C. Gennari, J. M. Goodman and I. Paterson, Tetrahedron: Asymmetry, 1995, 6, 2613–2636 CrossRef CAS .
5. (a) K.-M. Chen, G. E. Hardtmann, K. Prasad, O. Repič and M. J. Shapiro, Tetrahedron Lett., 1987, 28, 155–158 CrossRef CAS ; (b) D. A. Evans, K. T. Chapman and E. M. Carreira, J. Am. Chem. Soc., 1988, 110, 3560–3578 CrossRef CAS ; (c) D. A. Evans and A. H. Hoveyda, J. Am. Chem. Soc., 1990, 112, 6447–6449 CrossRef CAS .
6. (a) C. J. Cowden and I. Paterson, Organic Reactions, ed. L. A. Paquette, Wiley, vol. 51, pp. 1–200, 1997 Search PubMed ; (b) I. Paterson, V. A. Doughty, G. J. Florence, K. Gerlach, M. D. McLeod, J. P. Scott and T. Trieselmann, in Organoboranes for Syntheses, ed. P. V. Ramachandran and H. C. Brown, American Chemical Society, ACS Symposium Series 783, ch. 14, pp. 195–206, 2001 Search PubMed ; (c) I. Paterson, C. J. Cowden and D. J. Wallace, in Modern Carbonyl Chemistry, ed. J. Otera, Wiley-VCH, 2000, pp. 249–297 Search PubMed .
7. I. Paterson, J. M. Goodman, M. A. Lister, R. C. Schumann, C. K. McClure and R. D. Norcross, Tetrahedron, 1990, 46, 4663–4684 CrossRef CAS .
8. I. Paterson, J. M. Goodman and M. Isaka, Tetrahedron Lett., 1989, 30, 7121–7124 CrossRef CAS .
9. (a) I. Paterson, D. J. Wallace and S. M. Velázquez, Tetrahedron Lett., 1994, 35, 9083–9086 CrossRef CAS ; (b) I. Paterson and D. J. Wallace, Tetrahedron Lett., 1994, 35, 9087–9090 CrossRef CAS ; (c) I. Paterson, D. J. Wallace and C. J. Cowden, Synthesis, 1998, 639–652 CrossRef CAS .
10. (a) I. Paterson, M. A. Lister and C. K. McClure, Tetrahedron Lett., 1986, 27, 4787–4790 CrossRef CAS ; (b) I. Paterson and M. A. Lister, Tetrahedron Lett., 1988, 29, 585–588 CrossRef CAS ; (c) I. Paterson and J. M. Goodman, Tetrahedron Lett., 1989, 30, 997–1000 CrossRef CAS .
11. S. Carmely and Y. Kashman, Tetrahedron Lett., 1985, 26, 511–514 CrossRef CAS .
12. M. Kobayashi, J. Tanaka, T. Katori, M. Matsuura, M. Yamashita and I. Kitagawa, Chem. Pharm. Bull., 1990, 38, 2409–2418 CrossRef CAS .
13. M. Kobayashi, K. Kawazoe, T. Okamoto, T. Sasaki and I. Kitagawa, Chem. Pharm. Bull., 1994, 42, 19–26 CrossRef CAS PubMed .
14. M. R. Bubb, I. Spector, A. D. Bershadsky and E. D. Korn, J. Biol. Chem., 1995, 270, 3463–3466 CrossRef CAS PubMed .
15. (a) C. A. Bewley, N. D. Holland and D. J. Faulkner, Experientia, 1996, 52, 716–722 CrossRef CAS PubMed ; (b) J. Kobayashi and M. Ishibashi, Chem. Rev., 1993, 93, 1753–1769 CrossRef CAS .
16. For other completed total syntheses, see: (a) K. C. Nicolaou, K. Ajito, A. P. Patron, H. Khatuya, P. K. Richter and P. Bertinato, J. Am. Chem. Soc., 1996, 118, 3059–3060 CrossRef CAS ; (b) K. C. Nicolaou, A. P. Patron, K. Ajito, P. K. Richter, H. Khatuya, P. Bertinato, R. A. Miller and M. J. Tomaszewski, Chem. – Eur. J., 1996, 2, 847–868 CrossRef CAS ; (c) I. Shin, S. Hong and M. J. Krische, J. Am. Chem. Soc., 2016, 138, 14246–14249 CrossRef CAS PubMed .
17. (a) I. Paterson and J. D. Smith, Tetrahedron Lett., 1993, 34, 5351–5354 CrossRef CAS ; (b) I. Paterson and S. A. Osborne, Tetrahedron Lett., 1990, 31, 2213–2216 CrossRef CAS .
18. I. Paterson and J. D. Smith, J. Org. Chem., 1992, 57, 3261–3264 CrossRef CAS .
19. I. Paterson and J. G. Cumming, Tetrahedron Lett., 1992, 33, 2847–2850 CrossRef CAS .
20. I. Paterson, J. G. Cumming, J. D. Smith and R. A. Ward, Tetrahedron Lett., 1994, 35, 441–444 CrossRef CAS .
21. I. Paterson, J. D. Smith, R. A. Ward and J. G. Cumming, J. Am. Chem. Soc., 1994, 116, 2615–2616 CrossRef CAS .
22. I. Paterson, J. G. Cumming, J. D. Smith, R. A. Ward and K.-S. Yeung, Tetrahedron Lett., 1994, 35, 3405–3408 CrossRef CAS .
23. I. Paterson, K.-S. Yeung, R. A. Ward, J. G. Cumming and J. D. Smith, J. Am. Chem. Soc., 1994, 116, 9391–9392 CrossRef CAS .
24. (a) I. Paterson, J. G. Cumming, R. A. Ward and S. Lamboley, Tetrahedron, 1995, 51, 9393–9412 CrossRef CAS ; (b) I. Paterson, J. D. Smith and R. A. Ward, Tetrahedron, 1995, 51, 9413–9436 CrossRef CAS ; (c) I. Paterson, R. A. Ward, J. D. Smith, J. G. Cumming and K. S. Yeung, Tetrahedron, 1995, 51, 9437–9466 CrossRef CAS ; (d) I. Paterson, K.-S. Yeung, R. A. Ward, J. D. Smith, J. G. Cumming and S. Lamboley, Tetrahedron, 1995, 51, 9467–9486 CrossRef CAS .
25. (a) G. R. Pettit, Z. A. Chicacz, F. Gao, C. L. Herald, M. R. Boyd, J. M. Schmidt and J. N. A. Hooper, J. Org. Chem., 1993, 58, 1302–1304 CrossRef CAS ; (b) G. R. Pettit, Z. A. Cichacz, F. Gao, C. L. Herald and M. R. Boyd, J. Chem. Soc., Chem. Commun., 1993, 1166–1168 RSC .
26. (a) M. Kobayashi, S. Aoki, H. Sakai, K. Kawazoe, N. Kihara, T. Sasaki and I. Kitagawa, Tetrahedron Lett., 1993, 34, 2795–2798 CrossRef CAS ; (b) M. Kobayashi, S. Aoki, H. Sakai, N. Kihara, T. Sasaki and I. Kitagawa, Chem. Pharm. Bull., 1993, 41, 989–991 CrossRef CAS ; (c) M. Kobayashi, S. Aoki and I. Kitagawa, Tetrahedron Lett., 1994, 35, 1243–1246 CrossRef CAS ; (d) M. Kobayashi, S. Aoki, K. Gato and I. Kitagawa, Chem. Pharm. Bull., 1996, 44, 2142–2149 CrossRef CAS .
27. N. Fusetani, K. Shinoda and S. Matsunaga, J. Am. Chem. Soc., 1993, 115, 3977–3981 CrossRef CAS .
28. (a) D. A. Evans, P. J. Coleman and L. C. Dias, Angew. Chem., Int. Ed. Engl., 1997, 36, 2738–2741 CrossRef CAS ; (b) D. A. Evans, B. W. Trotter, B. Cǒté and P. J. Coleman, Angew. Chem., Int. Ed. Engl., 1997, 36, 2741–2744 CrossRef CAS ; (c) D. A. Evans, B. W. Trotter, B. Côté, P. J. Coleman, L. C. Dias and A. N. Tyler, Angew. Chem., Int. Ed. Engl., 1997, 36, 2744–2747 CrossRef CAS .
29. (a) J. Guo, K. J. Duffy, K. L. Stevens, P. I. Dalko, R. M. Roth, M. M. Hayward and Y. Kishi, Angew. Chem., Int. Ed., 1998, 37, 187–192 CrossRef CAS ; (b) M. M. Hayward, R. M. Roth, K. J. Duffy, P. I. Dalko, K. L. Stevens, J. Guo and Y. Kishi, Angew. Chem., Int. Ed., 1998, 37, 192–196 CrossRef CAS .
30. R. K. Pettit, S. C. McAllister, G. R. Pettit, C. L. Herald, J. M. Johnson and Z. A. Cichacz, Int. J. Antimicrob. Agents, 1998, 9, 147–152 CrossRef CAS .
31. G. Menchon, A. E. Prota, D. Lucena-Agell, P. Bucher, R. Jansen, H. Irschik, R. Müller, I. Paterson, J. F. Díaz, K.-H. Altmann and M. O. Steinmetz, Nat. Commun., 2018, 9, 2106 CrossRef .
32. R. Bai, G. F. Taylor, Z. A. Cichacz, C. L. Herald, J. A. Kepler, G. R. Pettit and E. Hamel, Biochemistry, 1995, 34, 9714–9721 CrossRef CAS PubMed .
33. For other completed total syntheses, see: (a) A. B. Smith III, Q. Y. Lin, V. A. Doughty, L. H. Zhuang, M. D. McBriar, J. K. Kerns, A. M. Boldi, N. Murase, W. H. Moser, C. S. Brook, C. S. Bennett, K. Nakayama, M. Sobukawa and R. E. Trout, Tetrahedron, 2009, 65, 6470–6488 CrossRef PubMed ; (b) A. B. Smith III, C. Sfouggatakis, C. A. Risatti, J. B. Sperry, W. Y. Zhu, V. A. Doughty, T. Tomioka, D. B. Gotchev, C. S. Bennett, S. Sakamoto, O. Atasoylu, S. Shirakami, D. Bauer, M. Takeuchi, J. Koyanagi and Y. Sakamoto, Tetrahedron, 2009, 65, 6489–6509 CrossRef PubMed ; (c) C. H. Heathcock, M. McLaughlin, J. Medina, J. L. Hubbs, G. A. Wallace, R. Scott, M. M. Claffey and C. J. Hayes, J. Am. Chem. Soc., 2003, 125, 12844–12849 CrossRef CAS PubMed ; (d) M. T. Crimmins, J. D. Katz, D. G. Washburn, S. P. Allwein and L. F. McAtee, J. Am. Chem. Soc., 2002, 124, 5661–5663 CrossRef CAS PubMed ; (e) M. Ball, M. J. Gaunt, D. F. Hook, A. S. Jessiman, S. Kawahara, P. Orsini, A. Scolaro, A. C. Talbot, H. R. Tanner, S. Yamanoi and S. V. Ley, Angew. Chem., Int. Ed., 2005, 44, 5433–5438 CrossRef CAS PubMed ; (f) L. M. Suen, M. A. Tekle-Smith, K. S. Williamson, J. R. Infantine, S. K. Reznik, P. S. Tanis, T. D. Casselman, D. L. Sackett and J. L. Leighton, Nat. Commun., 2018, 9, 4710 CrossRef PubMed .
34. I. Paterson, D. Y.-K. Chen, M. J. Coster, J. L. Acena, J. Bach, K. R. Gibson, L. E. Keown, R. M. Oballa, T. Trieselmann, D. J. Wallace, A. Hodgson and R. D. Norcross, Angew. Chem., Int. Ed., 2001, 40, 4055–4060 CrossRef CAS PubMed .
35. (a) I. Paterson and R. M. Oballa, Tetrahedron Lett., 1997, 38, 8241–8244 CrossRef CAS ; (b) I. Paterson, M. J. Coster, D. Y.-K. Chen, R. M. Oballa, D. J. Wallace and R. D. Norcross, Org. Biomol. Chem., 2005, 3, 2399–2409 RSC .
36. (a) I. Paterson, K. R. Gibson and R. M. Oballa, Tetrahedron Lett., 1996, 37, 8585–8588 CrossRef CAS ; (b) I. Paterson and L. A. Collett, Tetrahedron Lett., 2001, 42, 1187–1191 CrossRef CAS ; (c) D. A. Evans, P. J. Coleman and B. Côté, J. Org. Chem., 1997, 62, 788–789 CrossRef CAS ; (d) D. A. Evans, B. Côté, P. J. Coleman and B. T. Connell, J. Am. Chem. Soc., 2003, 125, 10893–10898 CrossRef CAS .
37. (a) I. Paterson, D. J. Wallace and K. R. Gibson, Tetrahedron Lett., 1997, 38, 8911–8914 CrossRef CAS ; (b) I. Paterson and M. J. Coster, Tetrahedron Lett., 2002, 43, 3285–3289 CrossRef CAS ; (c) I. Paterson, M. J. Coster, D. Y. K. Chen, K. R. Gibson and D. J. Wallace, Org. Biomol. Chem., 2005, 3, 2410–2419 RSC .
38. I. Paterson, M. J. Coster, D. Y. K. Chen, J. L. Aceña, J. Bach, L. E. Keown and T. Trieselmann, Org. Biomol. Chem., 2005, 3, 2420–2430 RSC .
39. H. C. Kolb, M. S. VanNieuwenhze and K. B. Sharpless, Chem. Rev., 1994, 94, 2483–2547 CrossRef CAS .
40. I. Paterson, K.-S. Yeung and J. B. Smaill, Synlett, 1993, 774–776 CrossRef CAS .
41. I. Paterson, D. Y. K. Chen, M. J. Coster, J. L. Aceña, J. Bach and D. J. Wallace, Org. Biomol. Chem., 2005, 3, 2431–2440 RSC .
42. D. A. Evans, B. W. Trotter, P. J. Coleman, B. Côté, L. C. Dias, H. A. Rajapakse and A. N. Tyler, Tetrahedron, 1999, 55, 8671–8726 CrossRef CAS .
43. I. Paterson, J. L. Acena, J. Bach, D. Y.-K. Chen and M. J. Coster, Chem. Commun., 2003, 462–463 RSC .
44. D. E. Williams, M. Roberge, R. Van Soest and R. J. Andersen, J. Am. Chem. Soc., 2003, 125, 5296–5297 CrossRef CAS PubMed .
45. D. E. Williams, M. Lapawa, X. Feng, T. Tarling, M. Roberge and R. J. Andersen, Org. Lett., 2004, 6, 2607–2610 CrossRef CAS PubMed .
46. (a) I. Paterson, E. A. Anderson, S. M. Dalby, J. H. Lim, J. Genovino, P. Maltas and C. Moessner, Angew. Chem., Int. Ed., 2008, 47, 3016–3020 CrossRef CAS PubMed ; (b) I. Paterson, E. A. Anderson, S. M. Dalby, J. H. Lim, J. Genovino, P. Maltas and C. Moessner, Angew. Chem., Int. Ed., 2008, 47, 3021–3025 CrossRef CAS ; (c) I. Paterson, P. Maltas and E. A. Anderson, Pure Appl. Chem., 2013, 85, 1133–1147 CAS .
47. I. Paterson, P. Maltas, S. M. Dalby, J. H. Lim and E. A. Anderson, Angew. Chem., Int. Ed., 2012, 51, 2749–2753 CrossRef CAS .
48. For other completed total syntheses, see: (a) G. W. O’Neil, J. Ceccon, S. Benson, M.-P. Collin, B. Fasching and A. Fürstner, Angew. Chem., Int. Ed., 2009, 48, 9940–9945 CrossRef ; (b) S. Benson, M.-P. Collin, G. W. O’Neil, J. Ceccon, B. Fasching, M. D. B. Fenster, C. Godbout, K. Radkowski, R. Goddard and A. Fürstner, Angew. Chem., Int. Ed., 2009, 48, 9946–9950 CrossRef CAS ; (c) S. Benson, M.-P. Collin, A. Arlt, B. Gabor, R. Goddard and A. Fürstner, Angew. Chem., Int. Ed., 2011, 50, 8739–8744 CrossRef CAS PubMed .
49. I. Paterson, E. A. Anderson, S. M. Dalby, J. H. Lim, P. Maltas, O. Loiseleur, J. Genovino and C. Moessner, Org. Biomol. Chem., 2012, 10, 5861–5872 RSC .
50. I. Paterson, E. A. Anderson, S. M. Dalby, J. H. Lim and P. Maltas, Org. Biomol. Chem., 2012, 10, 5873–5886 RSC .
51. I. Paterson, E. A. Anderson, S. M. Dalby and O. Loiseleur, Org. Lett., 2005, 7, 4121–4124 CrossRef CAS .
52. I. Paterson, E. A. Anderson, S. M. Dalby and O. Loiseleur, Org. Lett., 2005, 7, 4125–4128 CrossRef CAS PubMed .
53. I. Paterson, E. A. Anderson and S. M. Dalby, Synthesis, 2005, 3225–3228 CrossRef CAS .
54. I. Paterson, E. A. Anderson, S. M. Dalby, J. H. Lim, P. Maltas and C. Moessner, Chem. Commun., 2006, 4186–4188 RSC .
55. I. Paterson, E. A. Anderson, S. M. Dalby, J. Genovino, J. H. Lim and C. Moessner, Chem. Commun., 2007, 1852–1854 RSC .
56. S. Hu, S. Jayaraman and A. C. Oehlschlager, J. Org. Chem., 1996, 61, 7513–7520 CrossRef CAS .
57. I. Paterson and S. M. Dalby, Nat. Prod. Rep., 2009, 26, 865–873 RSC .
58. A. E. Wright, J. K. Reed, J. Roberts and R. E. Longley, US Pat., US2008033035Chem. Abstr., 2008, 148, 230103 Search PubMed .
59. I. Paterson, S. M. Dalby, J. C. Roberts, G. J. Naylor, E. A. Guzmán, R. Isbrucker, T. P. Pitts, P. Linley, D. Divlianska, J. K. Reed and A. E. Wright, Angew. Chem., Int. Ed., 2011, 50, 3219–3223 CrossRef CAS PubMed .
60. I. Paterson, T. Paquet and S. M. Dalby, Org. Lett., 2011, 13, 4398–4401 CrossRef CAS .
61. I. Paterson and S. Williams, Isr. J. Chem., 2017, 57, 192–201 CrossRef CAS .
62. (a) J. Willwacher, N. Kausch-Busies and A. Fürstner, Angew. Chem., Int. Ed., 2012, 51, 12041–12046 CrossRef CAS PubMed ; (b) D. Mailhol, J. Willwacher, N. Kausch-Busies, E. E. Rubitski, Z. Sobol, M. Schuler, M. H. Lam, S. Musto, F. Loganzo, A. Maderna and A. Fürstner, J. Am. Chem. Soc., 2014, 136, 15719–15729 CrossRef CAS PubMed .
63. I. Paterson, K. K. H. Ng, S. Williams, D. C. Millican and S. M. Dalby, Angew. Chem., Int. Ed., 2014, 53, 2692–2695 CrossRef CAS PubMed .
64. R. Jansen, H. Irschik, H. Reichenbach and G. Höfle, Liebigs Ann., 1997, 1725–1732 CrossRef CAS .
65. D. Janssen, D. Albert, R. Jansen, R. Müller and M. Kalesse, Angew. Chem., Int. Ed., 2007, 46, 4898–4901 CrossRef CAS .
66. T. Brodmann, D. Janssen and M. Kalesse, J. Am. Chem. Soc., 2010, 132, 13610–13611 CrossRef CAS .
67. S. Williams, J. Jin, S. B. J. Kan, M. Li, L. J. Gibson and I. Paterson, Angew. Chem., Int. Ed., 2017, 56, 645–649 CrossRef CAS PubMed .
68. I. Paterson, S. B. J. Kan and L. J. Gibson, Org. Lett., 2010, 12, 3724–3727 CrossRef CAS PubMed .
69. S. I. Shirokawa, M. Kamiyama, T. Nakamura, M. Okada, A. Nakazaki, S. Hosokawa and S. Kobayashi, J. Am. Chem. Soc., 2004, 126, 13604–13605 CrossRef CAS PubMed .
70. J. Uenishi, R. Kawahama, O. Yonemitsu and J. Tsuji, J. Org. Chem., 1998, 63, 8965–8975 CrossRef CAS .
71. G. Stork and K. Zhao, Tetrahedron Lett., 1989, 30, 2173–2174 CrossRef CAS .
72. I. Paterson, L. J. Gibson and S. B. J. Kan, Org. Lett., 2010, 12, 5530–5533 CrossRef CAS PubMed .
73. W. C. Still and C. Gennari, Tetrahedron Lett., 1983, 24, 4405–4408 CrossRef CAS .
74. Y. Inahashi, M. Iwatsuki, A. Ishiyama, A. Matsumoto, T. Hirose, J. Oshita, T. Sunazuka, W. Panbangred, Y. Takahashi, M. Kaiser, K. Otoguro and S. Omura, Org. Lett., 2015, 17, 864–867 CrossRef CAS PubMed .
75. M. J. Anketell, T. M. Sharrock and I. Paterson, Angew. Chem., Int. Ed., 2020, 59, 1572–1576 CrossRef CAS PubMed .
76. D. Seebach, R. Naef and G. Calderari, Tetrahedron, 1984, 40, 1313–1324 CrossRef .
77. (a) I. Paterson and A. N. Hulme, Tetrahedron Lett., 1990, 31, 7513–7516 CrossRef CAS ; (b) I. Paterson and A. N. Hulme, J. Org. Chem., 1995, 60, 3288–3300 CrossRef CAS .
78. J. G. Solsona, J. Nebot, P. Romea and F. Urpí, J. Org. Chem., 2005, 70, 6533–6536 CrossRef CAS PubMed .
79. J. Esteve, S. Matas, M. Pellicena, J. Velasco, P. Romea, F. Urpí and M. Font-Bardia, Eur. J. Org. Chem., 2010, 3146–3151 CrossRef CAS .
80. R. Nagase, Y. Oguni, T. Misaki and Y. Tanabe, Synthesis, 2006, 3915–3917 CAS .
81. C. Lecourt, S. Dhambri, L. Allievi, Y. Sanogo, N. Zeghbib, R. B. Othman, M.-I. Lannou, G. Sorin and J. Ardisson, Nat. Prod. Rep., 2018, 35, 105–124 RSC .
82. M. J. Anketell, T. M. Sharrock and I. Paterson, Org. Biomol. Chem., 2020, 18, 8109–8118 RSC .
83. (a) S. J. Mickel, D. Niederer, R. Daeffler, A. Os,ani, E. Kuesters, E. Schmid, K. Schaer, R. Gamboni, W. Chen, E. Loeser, F. R. Kinder, K. Konigsberger, K. Prasad, T. M. Ramsey, O. Repic, R.-M. Wang, G. Florence, I. Lyothier and I. Paterson, Org. Process Res. Dev., 2004, 122–130 CrossRef CAS ; (b) G. J. Florence, N. M. Gardner and I. Paterson, Nat. Prod. Rep., 2008, 25, 342–375 RSC .

---

Footnote

† Authors contributed to this manuscript equally.

This journal is © The Royal Society of Chemistry 2021