Synthesis of Waixenicin A: Exploring Strategies for Nine-Membered Ring Formation

Abstract

We present a comprehensive account on our efforts behind the recently published synthesis of waixenicin A. Our approach for constructing the dihydropyran ring relied on an Achmatowicz rearrangement. For the assembly of the nine-membered ring four distinct strategies were investigated. Our initial attempts using radical-based addition/fragmentation reactions targeting the C7-C11 bond proved unsuitable for accessing the 6/9-bicycle. By employing anionic fragmentation conditions at the furfuryl alcohol stage, we successfully reached a 5/9-bicycle. However, subsequent ring-expansion was unsuccessful. Alternative approaches, such as Nozaki–Hiyama–Kishi or Heck reactions to connect the C6-C7 bond, also encountered difficulties, with no nine-membered ring formation observed. Our first breakthrough came from our attempts to install the C5-C6 bond via an intramolecular alkylation. Surprisingly, subsequent functional group modifications proved unexpectedly challenging, necessitating a redesign of our synthetic route. Drawing from all our investigations, we concluded that construction of the C9-C10 bond would enable efficient nine-membered ring alkylation and would facilitate the installation of the desired substitution pattern along the southern periphery. Exploration of this strategy yielded further surprises but ultimately led to the successful synthesis of waixenicin A and 9-deacetoxy-14,15-deepoxyxeniculin. For the latter compound, a bioinspired one-step rearrangement to xeniafauranol A was achieved.

Introduction

The xenicanes are a group of marine diterpenoid natural products, which are produced by soft corals of the Xenia genus. After the isolation of the first member in 1977,¹ a wide variety of different congeners have been isolated over the years.² Xenicanes possess unique bicyclic skeletons, consisting of a nine-membered carbocycle, connected to a smaller ring of varying size. The size of the smaller ring allows these secondary metabolites to be differentiated into further sub families. Members of the xenicin subclass possess a 11-oxabicyclo[7.4.0]tridecane ring system, with the (partially) reduced pyran ring containing an acetal moiety. Notable members of the xenicin subgroup include waixenicin A (1),³ waixenicin B (2), xenicin (3),¹ 9-deacetoxy-14,15-deepoxyxeniculin (4),⁴ cristaxenicin A (5),⁵ as well as havannahine (6) (Figure 1).⁶ The large structural diversity within the different xenicane subclasses arises from the varying degree of oxidation along the nine-membered ring as well as the northern side chain. Many congeners were found to exhibit interesting biological activities.⁷ Waixenicin A (1), isolated in 1984 from Sarcothelia edmondsoni by Clardy and Scheuer was found to act as a very potent inhibitor of transient receptor potential melastatin 7 (TRPM7) channels and is able to inhibit cell proliferation in the nanomolar range (Mg²⁺-dependent IC₅₀ =16 nM).⁸ In addition, this activity was found to be highly selective, with no observed activity against TRPM6, the closest homologue to TRPM7. These properties render 1 a highly promising lead compound for the development of new agents targeting brain diseases or cancer.⁹ Due to these biological properties as well as its intriguing structure, we became interested to develop a synthesis of 1 and lay the foundation to access structurally related natural products.

Figure 1. Selected xenicins natural products.

Prior to our work, other members of the xenicanes have already attracted the attention of synthetic chemists.¹⁰ Despite the early discovery of xenicin (3), the first member, in 1977, it took until 2000 for the synthesis of a xenicane to be reported.¹¹ Leumann’s synthesis of coraxeniolide A was thereby based on early work of Pfander¹² as well as of Corey¹³ and featured a Grob-fragmentation to forge the cyclononene core of the natural product. In 2008, Corey reported an elegant second synthesis of coraxeniolide A, employing the same strategy for the ring construction.¹⁴ Other strategies successfully applied in the synthesis of xenicanes are the intramolecular Tsuji-Trost reaction as featured in Corey’s synthesis of antheliolide A,¹⁵ the ring-closing metathesis applied by Altmann to access blumiolide C,¹⁶ the Suzuki–Miyaura coupling Williams used to forge the nine-membered ring of 4-hydroxydictyolactone,¹⁷ as well as the Nozaki-Hiyama-Kishi strategy employed in Altmann’s recent synthesis of isoxeniolide A.¹⁸ Interestingly, the synthesis of a member of the xenicin subclass has remained unsolved. However, there are two total syntheses of closely related natural products (Scheme 1).

Scheme 1. A) Hosokawa’s synthesis of alcyonolide (7); B) Yao’s synthesis of plumisclerin A (11); C) Oxidation of deoxyxeniolide B (14).

In 2022, Hosokawa reported the synthesis of the seco-xenicin alcyonolide (7) for which the nine-membered ring is disconnected at the C6-C7 bond (Scheme 1A).¹⁹ Their strategy borrowed from their previous work towards the core structure of cristaxenicin A (5).²⁰ A hetero-Diels–Alder reaction between 8 and 9 allowed them to forge the dihydropyran-motif of the natural product at an early stage. Ten additional steps were required for the installation of the C11- and C12-side chains of 7. Plumisclerin A (11) is another natural product closely related to the xenicins and possesses a unique tricyclo[4.3.1.0^1,5]decane ring system (Scheme 1B). As 11 has been isolated from the same organism as cristaxenicin A (5), it is thought to be biosynthetically derived from an intramolecular [2+2]-cycloaddition of 5.²¹ In 2018, Yao reported a synthesis of 11 involving the conversion of 12 to 13 via an intramolecular SmI₂ mediated ketyl radical addition to an α,β-unsaturated ester to furnish the tricyclic core.²² The installation of the dihydropyran and of the C12-side chain was accomplished in 14 further steps. While both molecules possess the pyran-moiety characteristic for the xenicins, they lack the strained nine-membered carbocycle. In many xenicanes, the inherent ring strain is further enhanced by the presence of a trisubstituted (E) - configured C7-C8 alkene. Experimental support for the resulting high reactivity of the alkene and its tendency to undergo oxidation was described by Higuchi (Scheme 1C).²³ Under atmospheric oxygen, a solution of deoxyxeniolide B (14) in chloroform underwent conversion to epoxides 15 and 16. Turner showed that cyclononenes favor the (Z) - over (E)-alkene by 2.87 kcal/mol.²⁴ We therefore considered strategies that involve stereoselective formation of the C7-C8 alkene or strongly oxidizing conditions in the presence of the alkene to be of high risk.

Results and Discussion

As shown in Figure 2A, we began our studies with the analysis of the key structural features of waixenicin A (1). As a member of the xenicin subclass, 1 features an 11-oxabicyclo[7.4.0]tridecane ring system and bears a total of four stereogenic centers. In addition to the endocyclic trisubstituted alkene, the characteristic nine-membered ring also bears an exo-methylene unit. The dihydropyran ring of the natural product contains an enol ether as well as an acetal moiety, both of which were envisioned to be acid labile. Additionally, 1 possesses a total of three delicate acetates. The formation of the strained cyclononene was identified as the main synthetic challenge and thus guided our retrosynthetic analyses as shown in Figure 2B.

Figure 2. Structural features of waixenicin A (1) and investigated key-steps.

Our early strategies connected the nine-membered ring between C7 and C11, leading to fragmentation-based approaches. We then investigated a series of approaches dissecting the nine-membered ring between C6 and C7 (Heck- and Nozaki–Hiyama–Kishi (NHK)-approach) as well as between C5 and C6 (radical and Barbier-addition, intramolecular alkylation). With our lessons learned along the way, we then entered a final alkylation approach dissecting C9 and C10. To access the dihydropyran motif, we identified building blocks arising from Achmatowicz reaction of furfuryl alcohols as ideal precursors and therefore conserved this disconnection in all of our investigated approaches.²⁵ With one of the acetates located on the dihydropyran subunit – installed at an early stage in most of our envisioned strategies – we planned to replace it with a more stable p-methoxybenzyl (PMB) protecting group.

(A). Ring construction via fragmentation reactions (C7-C11 connection)

Fragmentations have emerged as a powerful tool for the construction of medium sized rings and have been used with great success by Corey¹⁴ and Leuman¹¹ to forge the cyclononene-ring of the xenicane coraxeniolide A. We initially focused on similar approaches, connecting the C7-C11 bond in the nine-membered ring. We first removed the northern side chain from 1, leading to literature-known aldehyde 17²⁶ and 6/9-ring system 18, which should be reductively coupled. Triflate 18 should be accessible from 19, which itself arises from a unique samarium diiodide-mediated radical cyclization/fragmentation cascade of enone 20 (Scheme 2). We planned to access this cyclization precursor (20) via a B-alkyl Suzuki-Miyaura cross-coupling between vinyl iodide 21 and alkene 22. Both building blocks can be traced back to 23 and 24 as readily available commercial starting materials.

Scheme 2. Retrosynthesis of the radical cyclization/fragmentation approach.

The envisioned key step was inspired by work of Molander²⁷ and should proceed via a SmI₂-mediated single-electron transfer (SET) to 20 furnishing ketyl radical 25 (Scheme 2B). A 6-endo-trig cyclization as depicted for 25 and reaction with an additional equivalent of SmI₂ should then give the 6/6/5-ring system 26. The latter should be poised to undergo a stereospecific Grob fragmentation forging the (E)-cylononene 19. The (E)-configuration of the trisubstituted double bond, would thereby be controlled by the syn-periplanar orientation between the C7-methyl group and the leaving group.²⁸

We began our endeavor with the preparation of iodo enone building block 21 (Scheme 3). Furfuryl alcohol 23 was subjected to a one-pot Achmatowicz reaction/acetylation procedure yielding racemic acetate 27. Enzymatic resolution with an immobilized Amano Lipase PS delivered enantioenriched 27.²⁹ Stereoretentive Tsuji–Trost reaction with p-methoxybenzyl alcohol (PMBOH) then gave the protected pyranone 28.³⁰ Treatment with iodine and pyridine delivered 21. Several Suzuki– Miyaura attempts with enone 21 led to decomposition, prompting us to convert the ketone into a silyl enol ether. Unexpectedly, treatment with lithium diisopropyl amide (LDA) and tert-butyldimethylsilyl trifluoromethanesulfonate (TBSOTf) did not result in the formation of the envisioned enol ether but delivered silyl protected alcohol 29. This nitrogen analog of the Meerwein–Ponndorf-Verley reduction has previously been described by Kowalski for certain enolizable α-alkoxy ketones.³¹ Going forward, we decided to perform the coupling with the reduced product 29.

Scheme 3. Synthesis of the radical cascade substrate.

The synthesis of the required alkene coupling partner 22 commenced with enantioselective reduction of 2-methyl-2-cyclopenten-1-one (24) using a Corey-Bakshi-Shibata (CBS) catalyst, delivering the corresponding allylic alcohol in high enantiopurity (95% ee).³² A directed Prilezhaev-epoxidation with m-chloroperbenzoic acid (m-CPBA) and Dess-Martin Periodinane (DMP) oxidation gave ketone 30. The quaternary carbon stereocenter of the envisioned building block was installed via a diastereoselective 1,2-addition of vinyl magnesium bromide followed by a boron trifluoride-diethyl etherate induced semi-pinacol rearrangement.³³ Silyl protection and enol ether formation provided 22. This compound set the stage for the coupling with vinyl iodide 29. Hydroboration of 22 and subsequent cross-coupling utilizing SPhos Pd G2 as catalyst delivered the desired product 31 in high yields. Having accomplished the coupling of both building blocks, we performed a global desilylation with tetra-n-butylammonium fluoride (TBAF) and allylic oxidation (MnO₂) to obtain ketone 32. From there, we accessed mesylate 20 as the substrate for the envisioned fragmentation in one step. Unfortunately, treating 20 with SmI₂ at different temperatures only resulted in complex reaction mixtures. Considering these outcomes, we explored a sequential approach, synthesizing silyl protected alcohol 33 to distinctly separate the coupling and fragmentation stages. Notably, 33 is incapable of undergoing the fragmentation reaction.

Initially, we repeated the conditions employed for the envisioned cascade of 20. However, using SmI₂ again led to the formation of a complex mixture of products (Table 1, entry 1). Going forward, we wanted to tune the reactivity of the employed samarium reductant by in situ formation of different samarium halides.³⁴ These are readily accessible by treating a solution of SmI₂ in tetrahydrofuran with a large excess of either lithium bromide or lithium chloride. Unfortunately, exposing 33 to a freshly prepared solution of SmBr2 also only provided a complex product mixture (entry 2). Employing SmCl2, which was prepared in a similar fashion, and tert-butanol (t-BuOH) as solvent, resulted in no conversion but recovery of unreacted starting material (entry 3). Addition of hexamethyl phosphoramide (HMPA) and hexafluoroisopropanol (HFIP) once again delivered a complex mixture (entry 4).

Table 1. Screening of the radical conjugate addition,

Entry	Additive	T (°C)	t (min)	Observation
1	-	-78	5	complex mixture
2	LiBr (6 equiv/SmI2)	-78	5	complex mixture
3	LiCl (6 equiv/SmI2), t-BuOH	-78	15	no conversion.
4	HMPA (10 equiv), HFIP (2 equiv)	-78	0.5	complex mixture

As the initial strategy proved unsuccessful, we started to explore a modified fragmentation-based strategy. Following a similar late-stage disconnection, we traced back waixenicin A (1) to intermediate 35, which should arise from a radical fragmentation of 36 (Scheme 4).³⁵ Having met challenges in constructing the required 6/6/5-ring system in our previous approach, we refined our strategy. Therefore, we planned the assembly of tricycle 36 via an intramolecular aldol condensation of aldehyde 37. The reduction of the resulting residual alkene would thereby most likely result in a cis-fusion of the two six-membered rings, making a late-stage epimerization necessary. Aldehyde 37 should be accessible via a Mukaiyama-Michael addition between 38 and the previously synthesized enone 28. Silyl ketene acetal 38 was further traced back to ketone 39.³⁶

Scheme 4. Retrosynthesis via a Mukaiyama-Michael addition/aldol sequence and radical fragmentation.

We started the synthesis of the required silyl ketene acetal 38 by protecting the alcohol and ketone functionalities as silyl (enol) ethers (Scheme 5). A hydroboration/oxidation protocol converted the vinyl group into alcohol 40. Protecting the new alcohol with another triisopropylsilyl group (Et₃N, TIPSOTf) and chemoselective cleavage of the enol ether (TBAF) delivered ketone 41 in high yield over two steps. Next, 41 was converted into the C₁-elongated aldehyde by epoxide formation through a Corey-Chaykowsky reaction and a Lewis acid-mediated Meinwald rearrangement for which MABR (= methylaluminum bis(4-bromo-2,6-di-tert-butylphenoxide) proved to be most effective.³⁷ The reaction produced a separable 1.4:1 mixture of diastereomeric aldehydes. Treatment of the undesired diastereomer with 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) delivered additional amounts of the desired isomer 42, which was obtained in 78% over two steps. Next, 42 was converted into silyl ketene acetal rac-38, via Pinnick-Lindgren-Kraus oxidation, methyl ester formation (MeI, DBU) and exposure to LDA and trimethylsilyl chloride (TMSCl).

Scheme 5. Synthesis of the 6/6/5-ring system.

With rac-38 in hand, we attempted the envisioned Mukaiyama-Michael addition to enone rac-28. We were pleased to see that boron trifluoride diethyl etherate was able to induce a clean diastereoselective addition, giving bicycle 43. Chemoselective deprotection of the primary TIPS ether using triethylamine trihydrofluoride and subsequent oxidation of the primary alcohol delivered aldehyde 37. At this stage, we were able to crystallize 37 and confirm its relative stereochemistry.³⁸ Exposure to proline promoted an intramolecular aldol condensation, furnishing the 6/6/5-ring system 44 in 60% yield. Hydrogenolytic removal of the residual alkene produced a cis-fusion between the two six-membered rings. Treatment with tetrabutylammonium fluoride (TBAF) delivered free alcohol 45, and reaction with 1,1'-thiocarbonyldiimidazole (TCDI) gave fragmentation precursor 36.

With 36 in hand, we started to investigate the fragmentation towards the nine-membered ring under radical conditions (Table 2). The initial approach involved premixing 36 with an excess of tributyltin hydride and azobisisobutyronitrile (AIBN) and heating the resulting mixture to 80 °C in benzene.

Table 2. Attempted radical fragmentation.

Entry	Conditions	Outcome
1a	n-Bu3SnH (4 equiv), AIBN (0.2 equiv), PhH, 80 °C	46 (66%)
2b	n-Bu3SnH (1.2 equiv), AIBN (0.2 equiv), PhH, 80 °C	47 (56%)
3b	n-Bu3SnH (1.2 equiv), AIBN (0.2 equiv), PhCl, 122 °C	48 (94%)
4b	TTMS (4 equiv), AIBN (0.2 equiv), PhH, 80 °C	47 (74%)

[a] premixing of n-Bu₃SnH/TTMS and AIBN; [b] slow addition of a solution of n-Bu₃SnH and AIBN to a heated solution of 36 over three hours.

This process yielded the mixed O,S-acetal 46 as the only reaction product (entry 1). To avoid this undesired side reaction, we next investigated the dropwise addition of a solution of tributyltin hydride and AIBN to a heated solution of thiocarbamate 36 (entry 2). Unfortunately, these conditions only resulted in the formation of Barton–McCombie deoxygenation product 47, which, despite formation of the secondary radical did not undergo the envisioned fragmentation. Trying to force the fragmentation, we then performed the reaction in chlorobenzene at 122 °C (entry 3). Intriguingly, under these conditions, we isolated lactone 48 (in 94% yield), whose structure was validated by single crystal X-ray analysis. We hypothesize that this product might be formed by a S_n2-type attack of the carbonyl oxygen of the ester at C8. In our final attempt, we replaced tributyltin hydride with tris(trimethylsilyl)silane (TTMS), a slower hydrogen donor (entry 4). However, much like the conditions described in entry 2, this approach also led to the formation of the deoxygenation product 47.

Given that none of the investigated radical fragmentation conditions yielded the desired nine-membered ring 35, we decided to return to approaches relying on an anionic Grob fragmentation (Scheme 6). In this approach, we chose to postpone the pyranone-forming Achmatowicz reaction to a later stage of the synthesis. By following a similar side chain introduction strategy as previously described, we traced the natural product back to pyranone 49, which we aimed to access via late-stage Achmatowicz reaction of the nine-membered ring fused furfuryl alcohol 50. The nine-membered ring was planned to be revealed via a Grob fragmentation of intermediate 51. The necessary fragmentation precursor, a 6/6/5-tricycle, was anticipated to be accessed by ring-closing metathesis and reduction from 52. Bicycle 52 can be dissected into simple building blocks, such as 3,4-dibromofuran 53 and ketone 54.

Scheme 6. Retrosynthesis via an anionic Grob fragmentation and RCM (ring closing metathesis).

Starting from freshly prepared 3,4-dibromofuran 53 (see SI for further information), we started the synthesis of the fragmentation precursor by mono-metalation (n-BuLi) and 1,2-addition to cyclopentanone 54, which is easily accessible through silyl protection of the previously used building block 39 (Scheme 7). Protecting the resulting tertiary alcohol (TMSCl, imidazole) delivered bromide 55, which was subjected to a second lithium-halogen exchange and trapping of the furyl-lithium species with dimethyl formamide. Subsequently, the resulting aldehyde was transformed to a vinyl group by Wittig olefination (n-BuLi, Ph₃PCH₃Br) furnishing diene 52. Ring-closing metathesis (RCM) with Grubbs second generation catalyst gave the 6/6/5-tricycle 56, which was exposed to hydrogenation conditions (Pd/C, H₂) to remove the alkene function. Formylation of the furan core by C-H metalation and reaction with dimethyl formamide then gave furfuraldehyde 57. From there, we proceeded to our envisioned fragmentation precursor 51 by global desilylation (TBAF) and mesylation of the secondary alcohol. Interestingly, this furfural derivative was reluctant to undergo a Grob fragmentation, showing no reaction upon treatment with one equivalent of sodium hydride or potassium bis(trimethylsilyl)amide and decomposition upon treatment with excess base. However, a reduction to yield alcohol 58 altered its reactivity and the obtained substrate underwent the desired Grob fragmentation to give cyclononenone 50. With this milestone achieved, we turned our attention towards the construction of the dihydropyrane-fragment of waixenicin A (1). Subjecting 50 to conditions commonly employed to induce the desired Achmatowicz rearrangement (NBS, NaOAc, NaHCO₃; VO(acac)₂, TBHP; ¹O₂)³⁹ we only observed decomposition of the starting material. We hypothesized that this is a result of the increased reactivity of the highly strained endocyclic (E)-alkene of the nine-membered ring, which enables different decomposition pathways.

Scheme 7. Successful Grob-fragmentation for the construction of the nine-membered ring.

To circumvent the unwanted decomposition, we investigated reversing the order of the fragmentation and the oxidation step. Reduction of furfural 57 with sodium borohydride delivered 60, which underwent smooth rearrangement to dihydro pyranone 61 in excellent yield and with good diastereoselectivity (d.r. = 7:1). The relative configuration of the obtained product was validated via single crystal X-ray analysis (Scheme 8).

Scheme 8. Attempted reduction of the pyranone.

For the reduction of the tetrasubstituted double bond in 61, we investigated varying lactol protecting groups (see SI for further information) and subjected those substrates to different reduction conditions. Unfortunately, even after a broad screen of homogenous, heterogenous, radical hydrogenation as well as 1,4-reduction protocols (e.g. Pd/C, H₂; [Ir(PCy₃)(py)(cod)]PF₆, H₂; Mn(dpm)₃, PhSiH₃; Cu(OAc)₂, dppf TMDS) we were not able to obtain the corresponding reduced products 62. We believe that this is a result of competing oxidopyrylium formation. Confronted with this problem we decided to abandon this approach and ultimately moved away from fragmentation-based strategies for the ring construction.

B). Ring construction via Heck or NHK-reaction (C6-C7 disconnection)

With the fragmentation strategies failing at the end, we set out to explore alternative strategies that hinge on the cyclization of acyclic precursors for the construction of the nine-membered ring. The next set of approaches dissected the carbocycle between C6 and C7 (Scheme 9). The first strategy aimed to close the nine-membered ring via a 9-endo Heck reaction. Due to reduced steric hindrance in the intermediate palladacycle, the 8-exo-trig cyclization mode is usually favoured in intramolecular Heck-reactions.⁴⁰ However, there are a few examples, which successfully utilized the 9-endo cyclization mode to access nine-membered carbocycles.⁴¹

Scheme 9. Retrosynthesis via intramolecular Heck cross-coupling.

As outlined in Scheme 9 and similar to our previous retrosyntheses, we envisioned a late installation of the northern side chain at C4, leading to bicycle 63. Due to incompatibility with the planned key step, we had to remove the triflate moiety in the six-membered ring, which served as a handle for side chain introduction in the previous approaches. Instead, we aimed for its installation via formylation of the enol ether with the Vilsmeier reagent 64, followed by α-selective addition of an allylic halide 65. Our strategy was to access bicycle 63 from vinyl dibromide 66, via Heck coupling, cross-coupling of the remaining vinyl bromide to install the missing methyl group and removal of the undesired double bond. Heck precursor 66, in turn, would arise from the three simple building blocks 28, 67 and 68 via 1,4-addition/Mukaiyama aldol and two olefinations.

Our studies began with the preparation of side chain 67, which was required for the envisioned conjugate addition (Scheme 10). Starting from literature-known vinyl bromide 69,⁴² bromide 67 was accessed by ester reduction (LAH) and triisopropylsilyl protection (TIPSOTf, Et₃N). Lithium-halogen exchange with t-BuLi and transmetalation delivered a cuprate which underwent smooth 1,4-addition to enantiopure enone 28. The resulting enolate was trapped with trimethylsilyl chloride (TMSCl) and furnished silyl enol ether 70, which was subjected to a ytterbium(III) triflate catalyzed Mukaiyama aldol reaction with aqueous formaldehyde. The reaction produced a complex mixture of the desired β-hydroxy ketone 72 together with the tentatively assigned formaldehyde over-addition product 71.

Scheme 10. Installation of the dibromoalkene and attempted cross-coupling.

The excess of formaldehyde was removed by heating the crude product mixture to 90 °C in the presence of 4 Å molecular sieves, to give gram-quantities of 72 in 58% yield over three steps.⁴³ Directed by the PMB protected acetal, this sequence was able to diastereoselectively install the C4a and C11a stereocenter of the natural product. We continued with silyl protection (TESCl, imH) of the primary alcohol and triflation of the ketone (PhNTf₂, LDA). The reduction of the obtained triflate (PdCl₂(PPh₃)₂, HCOOH) occurred alongside the removal of the recently installed silyl ether. Oxidation utilizing Dess-Martin Periodinane (DMP) delivered aldehyde 73. After standard olefination protocols failed to install the desired vinyl group, we had to rely on a cerium(III)-modified variant of the Peterson olefination.⁴⁴ Addition of 73 to a premixed suspension of cerium(III) chloride and (trimethylsilyl)methyllithium afforded a diastereomeric mixture of β-hydroxy silanes, which underwent elimination towards the vinyl group upon exposure to sodium bis(trimethylsilyl)amide. Removal of the TIPS group (TBAF), oxidation towards the corresponding aldehyde (DMP) and Ramirez-olefination (PPh₃, CBr₄) then yielded gem-dibromoalkene 66 as the Heck-precursor. Utilizing phosphine-free conditions, previously employed in the closure of nine-membered rings,⁴⁵ we were unable to obtain the desired product 74, since the substrate decomposed under the harsh reaction conditions.

Diverting from intermediate 66, we also wanted to explore an alternative approach (Scheme 11A). Applying the same end game disconnections as in the Heck approach, we traced waixenicin A (1) back to bicycle 75. The closure of the nine-membered ring was envisioned through an intramolecular Nozaki–Hiyama–Kishi (NHK) reaction of 76. Installation of the missing methyl group should again proceed via cross-coupling. Conveniently, aldehyde 76 could be easily derived from intermediate 66.

Scheme 11. Retrosynthesis via NHK reaction.

Since the direct installation of the required aldehyde functionality to 66 via Grubbs-Wacker oxidation was met with no success,⁴⁶ we choose to rely on a two-step procedure consisting of olefin cross metathesis with vinylboronic acid pinacol ester 77 and subsequent oxidative boron cleavage (NaBO₃·4H₂O) to get hands on cyclization precursor 76 (Scheme 11B). Standard NHK-conditions utilizing chromium(II) chloride and catalytic nickel(II) chloride however did not lead to the formation of carbocycle 75 and only unreacted 76 was recovered.⁴⁷

C). Ring construction via radical addition, Barbier-type addition or intramolecular alkylation (C5-C6 disconnection)

Next, we shifted our attention to a series of retrosynthetic strategies, which dissected the carbocycle between C5 and C6 (Scheme 12). We planned to obtain the respective cyclization precursors utilizing similar chemistry as for 66 and 76.

Scheme 12. Retrosynthesis targeting the C5-C6 bond.

Our goal was to access waixenicin A (1) from bicycle 63, which we traced back to three different linear precursors: enone 78 which should undergo a radical initiated cyclization onto the exocyclic-enone moiety to form the nine-membered ring, aldehyde 79 as precursor for an intramolecular Barbier-type addition and sulfone 80 for which we envisioned a cyclization via an intramolecular α-alkylation of the sulfone. All these precursors should be obtained by conjugate addition of metalated 81 to enone 28. This would again be followed by a Mukaiyama aldol reaction with formaldehyde (68) and straightforward follow-up transformations. First focusing on the radical addition approach, we had to get hands on the suitable side chain for the conjugate addition (Scheme 13).

Scheme 13. Attempted intramolecular radical cyclization.

Reduction of 69 with diisobutyl aluminum hydride (DIBAL-H) delivered an aldehyde, which was next converted to the corresponding conjugated methyl ester via Wittig reaction with phosphonium salt 82. Following the reduction towards the allylic alcohol (DIBAL-H) and silyl protection (TBSCl, imH), bromide 81 was obtained. Employing the previously established metalation protocol (Scheme 10), the formed vinyl cuprate underwent smooth addition to enone 28. The formed enolate was trapped as its enol ether (TMSCl). Ytterbium(III) triflate catalyzed Mukaiyama aldol reaction and subsequent removal of formaldehyde over-addition products furnished β-hydroxy ketone 83 in 47% over three steps. For the following dehydration towards the corresponding enone 84, conditions utilizing dicyclohexylcarbodiimide (DCC) in combination with substoichiometric amounts of copper(I) chloride proved to be most effective.⁴⁸ Cleavage of the silyl ether (Et₃N·3HF) and conversion of the resulting allylic alcohol into a chloride by mesylation (MsCl, Et₃N) and treatment with LiCl then gave radical cyclization precursor 78. An initial attempt to induce the planned radical cyclization by heating a premixed solution of enone 78, tributyltin hydride and AIBN only resulted in decomposition of the starting material. Trying to slowly add a solution of tributyltin hydride and AIBN to a heated solution of 78, also led to decomposition and trace amounts of the corresponding 1,4-reduction product.⁴⁹

For the synthesis of the Barbier-cyclization substrate, β-hydroxy ketone 83 served as synthetic starting point (Scheme 14). The primary alcohol was first silyl protected (TESCl, imH). Conversion of the ketone into the corresponding enol triflate (PhNTf2, LHMDS) and subsequent reduction (PdCl₂(PPh₃)₂, HCOOH) then delivered the enol ether function. Chemoselective deprotection of the TES-group with pyridinium p-toluenesulfonate (PPTS) gave the free alcohol 86.

Scheme 14. Synthesis of the Barbier cyclization precursor.

Acetylation (Ac₂O, DIPEA [(i-Pr)₂NEt]) and cleavage of the TBS group released the allylic alcohol, which was converted into allyl chloride 87 following the previously used mesylation/displacement strategy. Treatment with sodium methoxide removed the acetyl protecting group and oxidation (DMP) delivered the highly unstable aldehyde 79, which was used directly in the following cyclization trials.

To investigate the key step (Table 3), we initially focused on SmI₂-mediated conditions, which have previously been used with great success in the cyclization of halocarbonyl compounds.⁵⁰ However, treating 79 with SmI₂ at -78 °C only led to the recovery of unreacted starting material (entry 1). Repeating the reaction at 25 °C resulted in the decomposition of 79 (entry 2). We then switched to conditions, which made use of hexamethyl phosphoramide (HMPA) as an additive, resulting in comparable outcomes. While performing the reaction at -78 °C gave unreacted 79 (entry 3), decomposition was observed at 0 °C (entry 4). We next turned our attention to indium(0)-mediated procedures, which are known to deliver high α-selectivity in allylation reactions.⁵¹ Stirring 79 with indium metal in a heterogenous mixture of water and dichloromethane once again resulted in no reaction (entry 5). The same outcome was observed, when tetra-n-butylammonium iodide (TBAI) was added to the reaction mixture to achieve the in situ formation of a more reactive allylic iodide (entry 6). Finally, heating 79, sodium iodide, and tin(II) chloride resulted in isolation of a diastereomeric mixture of seven-membered ring 89 emerging from 7-endo-trig cyclization in 86% yield (entry 7).⁵² Unfortunately, the intramolecular allylation occurred with exclusive γ-selectivity and not even traces of the desired α-allylation product were detected.

Table 3. Screen of Barbier cyclization conditions.

Entry	Conditions	Solvent	T (°C)	Observation
1	SmI2	THF	-78	no reaction
2	SmI2	THF	25	decomposition
3	SmI2, HMPA	THF	-78	no reaction
4	SmI2, HMPA	THF	0	decomposition
5	In0	H2O, CH2Cl2	25	no reaction
6	In0, TBAI	H2O, CH2Cl2	25	no reaction
7	SnCl2, NaI	DMF	25 to 60	89 (86%)

Confronted with the selectivity issues in the Barbier-cyclizations, our focus shifted to the alkylation-based approach (Scheme 15). Early studies revealed that an alkylation precursor bearing an enol triflate in the nine-membered ring would decompose under the envisioned cyclization conditions. Consequently, we decided to proceed with our previous intermediate 86 where the triflate had been removed. The sulfone moiety was successfully installed through a two-step process. First, the primary alcohol was converted to an iodide via the Garegg-Samuelsson reaction. Second, the iodide was displaced with sodium phenyl sulfinate. Deprotection of the allylic alcohol (Et₃N·3HF) and mesylation/displacement gave chloride 80 and bromide 90 as cyclization precursors.

Scheme 15. Successful 9-ring alkylation.

Our next step was to test the intramolecular alkylation of compounds 80 and 90 (Table 4). Initially, encouraging results were obtained when LHMDS was added to a solution of 80 at 60 °C. Under these conditions, the cyclized product 91 was obtained in 10% yield (entry 1). KHMDS and 18-crown-6 led to starting material decomposition (entry 2). A similar outcome was observed when KHMDS was added to bromide 90 at 0 °C (entry 3). Much to our satisfaction, the use of NaHMDS successfully induced a clean cyclization of 90, delivering 91 as a single diastereomer with no traces of the undesired competing S_N2’ displacement (entry 4).

Table 4. Investigation of the intramolecular alkylation.

Entry	Substrate	condition	T (°C)	t (h)	Observation
1	80	LHMDS	60	6	10%
2	80	KHMDS, 18-crown-6	60	2	decomposition
3	90	KHMDS	0 to 25	1.25	decomposition
4	90	NaHMDS	0	1.25	67%

Having accomplished the synthesis of the 6/9-ring system, new problems quickly arose when we attempted removal of the sulfone moiety (Scheme 16). A broad range of commonly used desulfonylation conditions (e.g. Mg, SmI₂, Al(Hg), Na(Hg), Li naphthalenide) was screened, but this either resulted in no conversion or, under more forcing conditions, decomposition of substrate 91 (see SI for further information). Similarly, efforts to reduce the sulfone to sulfide 92 remained unsuccessful. As reductive conditions were met with failure, we next attempted to solve the problem by oxidizing the alkyl sulfone to ketone 93. However, stirring with lithium bis(trimethylsilyl)amide (LHMDS) and subsequent treatment with Vedejs reagent (MoOPH = oxodiperoxymolybdenum(pyridine)-(hexamethylphosphoric triamide) only gave unreacted starting material.⁵³ Additionally, we investigated a procedure reported by Uguen and Kishi, which should convert the sulfone into a trialkyl borane 94.⁵⁴ Attempted deprotonation by stirring with LHMDS and exposure to 9-BBN however only gave unreacted starting material. Switching to methyl lithium as a sterically less hindered base also resulted in the same outcome. To determine, whether this result was caused by 91 being reluctant to deprotonation, we then performed a hydrogen-deuterium exchange experiment. Stirring 91 with LHMDS for 45 minutes and addition of D2O thereby led to no deuterium incorporation, suggesting that the C5 proton might be shielded by the nine-membered ring and inaccessible for deprotonation. This assumption was confirmed by a close inspection of the three-dimensional structure of 91, obtained by 2D NMR analysis and computational optimization at the B3LYP-D3/6-31+G(d,p) level of theory.

Scheme 16. Attempted desulfonylation and deprotonation experiment.

Given that our envisioned strategies for sulfone removal proved unsuccessful, and efforts to develop a cyclization precursor with a more electron-deficient sulfone—which should have facilitated desulfonylation—also failed, we ultimately decided to abandon this approach.

D). Ring construction via intramolecular alkylation (C9-C10 disconnection)

Encouraged by the result of the previous alkylation, our recent synthesis of cornexistin⁵⁵ and Corey’s synthesis of the xenicane antheliolide A,¹⁵ we decided to explore a revised alkylation strategy (Scheme 17). To this end, waixenicin A (1) was traced back to bicycle 96, which featured an enol triflate as a handle for the late-stage introduction of the side chain via sequential coupling. The nine-membered ring was dissected between C9 and C10 leading to β-keto sulfone 97 or β-keto ester 98 as potential precursors for an intramolecular alkylation. Due to the presence of a β-ketone functionality, the removal of the activating group should be facilitated. Both precursors were further disconnected to reveal dithiane 99. As the previously conducted approaches revealed the conjugate addition and enolate functionalization as powerful method to construct the C4a and C11a stereocenters of the natural product, we aimed to assemble 99 using a similar strategy from 28, 100, and 101.

Scheme 17. Retrosynthesis via disconnection at C9-C10.

Our early studies revealed that the enolate functionalization with homoallylic side chain building blocks 100 and 103 was problematic. While we intended to perform the conjugate addition of 101 and trapping of the enolate in a one-pot procedure, both steps were independently investigated in our early studies. Deprotonation of 1,4-addition product 102 with LHMDS, KHMDS, NaHMDS or NaH and exposure of the resulting enolate to an excess of iodide 100 or mesylate 103 in tetrahydrofuran yielded discouraging initial results. The desired alkylation product 104 was not detected as the enolate decomposed over time. Upon warming of the reaction mixture, the unreacted electrophiles remained as the only defined compound in the crude reaction mixture (Table 5, entries 1–4). Trying to increase the reactivity of the enolate, we examined the addition of HMPA to deprotonated 102 and doubling of the employed equivalents of 100 (entry 5). However, these conditions, as well as employing a more reactive magnesium enolate - prepared by deprotonation with NaHMDS in the presence of magnesium chloride (entry 6) - again just resulted in the recovery of unreacted electrophile and decomposition of 102. Switching the solvent to dimethyl formamide (DMF) also led to mostly decomposed 102. But this time, we also observed the first encouraging signals in the crude NMR (entries 7-9). Conducting the reaction with more equivalents of the electrophile and at higher concentration, did not give any improvements (entry 10).⁵⁶

Table 5. Reaction screen employing homoallylic electrophiles 100 and 103.

Entry	base	electrophile	solvent	Observation
1	LHMDS	100 or 103 (2.5 equiv)	THF	decomposition
2	KHMDS	100 or 103 (2.5 equiv)	THF	decomposition
3	NaHMDS	100 or 103 (2.5 equiv)	THF	decomposition
4	NaH	100 or 103 (2.5 equiv)	THF	decomposition
5	LHMDS, HMPA (2 equiv)	100 (6 equiv)	THF	decomposition
6	NaHMDS, MgCl2 (2 equiv)	100 (3 equiv)	DMF	decomposition
7	LHMDS	100 (3 equiv)	DMF	decomposition
8	KHMDS	100 (3 equiv)	DMF	decomposition
9	NaHMDS	100 (3 equiv)	DMF	decomposition
10	LHMDS	100 (6 equiv)	DMF	decomposition

Therefore, we decided to move away from the homoallylic electrophiles which appeared to be insufficiently reactive. Performing the reaction with allyl iodide in the envisioned one-pot procedure, we were able to isolate product 105 in good yield (Scheme 18). HMPA proved to be a crucial co-solvent, as significant amounts of the undesired dithiane 1,2-addition product were isolated in its absence. Despite the relatively lengthy sequence required to convert the allyl chain into the envisioned side chain, we decided to proceed with 105 during the project's early stages.

Scheme 18. Installation of the C4a and C11a stereocenters and optimization.

Reasoning that the required sequence would rely on standard transformations and should not pose too many problems, we deemed this strategy appropriate to quickly obtain some advanced material to investigate the alkylation key step. We proceeded by converting ketone 105 into the corresponding triflate (LHMDS, PhNTf₂). A hydroboration/oxidation protocol followed by a subsequent Swern oxidation then delivered unstable aldehyde 106. Compound 106 was next converted into methyl ketone 107 via a 1,2-addition of methyl magnesium bromide and another Swern oxidation of the resulting inconsequential mixture of diastereomers. From this point, we accessed 99 with the full required C4a side chain through a Horner-Wadsworth-Emmons (HWE) reaction with phosphonate 108 proceeding in high yield and (E)-selectivity. The resulting ester 109 was reduced (DIBAL-H) to the allylic alcohol and then silyl-protected (TBSCl, imH). Later, we optimized this sequence by employing methyl vinyl ketone equivalent 110, already bearing the additional methyl group of the side chain, as well as a vinyl boronic acid pinacol ester as a masked ketone for the chain elongation via HWE reaction. The one-pot conjugate addition/alkylation sequence gave 111 in 51% yield. Triflation proceeded smoothly, and treatment with sodium perborate tetrahydrate (NaBO₃·4H₂O) delivered methyl ketone 107, intersecting the previous sequence.

With a robust access to dithiane 99 in hand, we first studied the β-keto sulfone containing cyclization precursor (Scheme 19). As methods relying on mercury salts or hypervalent iodine reagents failed, we unmasked the aldehyde by exposure to a large excess of methyl iodide and calcium carbonate. Due to the instability of the formed aldehyde, it was used without further purification in the following 1,2-addition with lithiated methyl phenyl sulfone delivering 112 in 73% over two steps. Cyclization precursor 97 was then accessed in three further steps through oxidation (DMP), desilylation (HF·pyridine), and Appel reaction (CBr₄, PPh₃). Exposure to potassium carbonate then effected a clean cyclization to give an inseparable 2:1 mixture of products 113 and 113’. Careful analysis of the 2D-NMR data for the obtained mixture revealed conformational isomerism with regards to the spatial orientation of the endocyclic alkene in 113 and 113’.⁵⁷ While the H8-proton in major compound 113 exhibits a strong NOESY-correlation with H11a, H8 shows a strong correlation with H4a in 113’. Commonly employed procedures for the desulfonylation of β-keto sulfones proved problematic once again. However, we found that exposure of 113/113’ to radical conditions employing tributyltin hydride and AIBN at elevated temperatures effected clean conversion to a single product 114. Unfortunately, NMR-data of the obtained product indicated that the desulfonylation was accompanied with the isomerization of the C7/C8 alkene. We hypothesize that this may result from the formation of an α-keto radical 115 at C10, which can undergo an intramolecular 3-exo-trig cyclization onto the C7/C8 alkene yielding an intermediate cyclopropyl radical 116. This radical would then undergo a rotation process leading to the thermodynamically more stable (Z)-alkene after reopening of the cyclopropane moiety.⁵⁸

Scheme 19. Successful cyclization and alkene isomerization.

Although we eventually found that treatment of 113/113’ with sodium hydrogentelluride (NaTeH) enabled anionic desulfonylation without concomitant isomerization, the reaction suffered from low yields (23%) and very poor scalability. This ultimately prompted us to investigate cyclization precursor 98, in which the sulfone is replaced by a trimethylsilyl ethyl (TMSE) ester, as this functionality can be easily removed under mild anionic conditions in the presence of an appropiate fluoride source (Scheme 20). Deprotection of dithiane 99 and reaction with deprotonated TMSE acetate delivered alcohol 117 as an inconsequential mixture of diastereomers. Oxidation to the β-keto ester was accomplished under Ley-conditions and a sequence of desilylation (HF·pyridine) and Appel reaction delivered the modified cyclization precursor 98. Potassium carbonate once again induced clean cyclization to deliver the nine-membered ring as a 1.8:1 mixture of conformational isomers. Treatment of the obtained crude product with tris(dimethylamino)sulfonium difluorotrimethylsilicate (TASF) at 23 °C resulted in decarboxylation giving conformational isomers 96 and 96’ in a 1.2:1 ratio (only the natural product-like conformation is shown). For the following methenylation, we had to rely on a Zr-based Takai-Lombardo protocol (Zn, CH₂I₂, PbCl₂, ZrCl₄), as other commonly employed procedures were met with failure.⁵⁹ The conversion of the ketone into an exo-methylene resulted in a drastic change in the ratio of the observed conformational isomers, with the natural product-like conformation becoming the dominant species (10:1 ratio). Having successfully synthesized the bicyclic ring system 18, we focused on introducing the side chain, for which we ultimately decided to rely on a stepwise build-up. One-carbon elongation was achieved using carbonylation conditions. The presence of the created push-pull system was crucial to effect the following deprotection of the p-methoxybenzyl (PMB) group without substrate decomposition. The resulting anomeric alcohols were obtained as a 1.8:1 mixture. Surprisingly, standard procedures failed to give anomeric acetates 118α/β, which were only formed when we applied a combination of Staab’s reagent (CDI), acetic acid, and catalytic DBU. Both acetates were separable by column chromatography allowing us to proceed with the desired epimer 118β and to recycle 118α by a deacetylation/acetylation sequence (see SI for further details). To install the side chain of waixenicin A (1) we then relied on a Brown allylation of 118β,⁶⁰ acetyl protection of the obtained major diastereomer, and a highly challenging olefin cross-metathesis between 119 and alkene 120,⁶¹ culminating in its first total synthesis.

Scheme 20. Final steps towards waixenicin A (1), 9-deacetoxy-14,15-deepoxyxeniculin (4) and xeniafaraunol A (122).

After this breakthrough, we turned our attention to the closely related natural product 9-deacetoxy-14,15-deepoxyxeniculin (4). The required simplified side chain was introduced by prenylation of 118β using boronate 121.⁶² Acetylation of the major diastereomer then gave 9-deacetoxy-14,15-deepoxyxeniculin (4). In addition, we demonstrated that 4 undergoes rapid rearrangement to xeniafaraunol A (122) upon exposure to potassium carbonate in methanol, raising questions about its biosynthetic origin.⁶³

Upon successfully accessing our targeted natural products, we began to explore strategies for enhancing the developed route (Scheme 21). Revisiting the early steps of the synthesis, we reasoned, that employing a thioorthoester instead of a dithiane in the initial 1,4-addition/trapping sequence would render the overall route more redox-efficient. This approach would enable us to circumvent the oxidation-step of 117, streamlining the synthetic process. Performing the initial step with tris(methylthio)methane, we were able to isolate an inseparable mixture of desired product 124 and 1,4-addition product 123. Subjecting this mixture to triflation conditions (KHMDS, PhNTf₂), we were able to separate desired triflate 125 from the obtained mixture of triflation products. The subsequent unmasking of the ketone and the ensuing HWE reaction proceeded smoothly, yielding 126. Reduction and silyl protection then gave 127. A quick and high-yielding unmasking of thioester 128 was achieved upon treatment of 127 with silver nitrate (92%). The addition of triethylamine to the reaction mixture was crucial to avoid loss of the TBS group. Claisen condensation with deprotonated TMSE acetate successfully intersected the previous route, giving β-keto ester 129 in 55% yield. Having shown that this strategy was generally able to improve our developed route, we aimed to improve the initial conjugate addition/trapping step. Surprisingly, even after extensive screening of different temperature profiles, additives and employed equivalents of iodide 110, we were unable to find conditions under which the alkylation of the 1,4-addition product went to completion. Regrettably, formation of substantial amounts of 123 alongside 124 hampered the overall efficiency of the new sequence.

Scheme 21. Optimization results and shortening of the synthesis.

In addition, we wanted to improve the endgame of the synthesis of 1. Ideally, we sought a strategy which would allow us to install all carbon atoms of the side chain in a single coupling event, thereby circumventing the problematic olefin-cross metathesis of the final step of the sequence. The asymmetric Suzuki reaction developed by Morken thereby caught our attention.⁶⁴ While the original report makes uses of aryl halides as coupling substrates, we were keen to find out, whether intermediate 18 could also serve as a substrate. The synthesis of the required geminal bis(boronate) side chain precursor began with isoprene monoxide (130). Treatment with titanium tetrachloride led to addition of a chloride and concurrent epoxide opening to yield a primary allylic alcohol.⁶⁵ PMB protection using Dudley’s PMB-reagent (131) then delivered allyl chloride 132.⁶⁶ Reaction with deprotonated bis[(pinacolato)boryl]methane (133) gave coupling partner 134 in good yield. Unfortunately, the coupling studies utilizing Josiphos-ligands in combination with palladium acetate only produced complex product mixtures and ultimately did not improve our end game sequence.

Conclusion

In conclusion, we have reported the development of the first synthetic entry to the previously inaccessible xenicin subclass of the xenicanes. Our early strategies relied on a ring-construction via fragmentation reactions. Although the envisioned radical addition/fragmentation cascade and radical fragmentation failed to forge the cyclononene core of the natural product, we successfully installed the 9-ring via a Grob-fragmentation. However, our attempts to convert the corresponding furfuryl alcohol into a pyranone were met with failure. Reversing the order of the fragmentation and the Achmatowicz reaction also proved fruitless, prompting us to explore two strategies aimed at connecting C6 and C7 via either a 9-endo-Heck-cross coupling or an NHK-reaction. As these strategies failed to create the nine-membered ring, we shifted our focus on a radical cyclization, a Barbier-cyclization and an intramolecular alkylation, which should form the ring via closure between C5-C6. While the radical addition failed to give a cyclized product, we only observed formation of a seven-membered ring under Barbier-conditions. A significant breakthrough was achieved when we investigated the alkylation approach, which successfully delivered the characteristic 6/9-ring system of this natural product class. However, the progress of this approach was hampered by our inability to remove the sulfone from the nine-membered ring. Encouraged by the successful alkylation, we examined a modified approach forming the latter carbocycle by connecting the carbons C9 and C10. This strategy ultimately led to the first total synthesis of waixenicin A (1). The successful route features a highly stereoselective conjugate addition/trapping sequence to forge the stereocenters at C4a and C11a. The use of a vinyl boronic acid ester as a methyl vinyl ketone equivalent facilitated the construction of the required C4a-side chain via sequential elongation. The utilization of a β-keto sulfone as cyclization substrate again proved problematic due to an unexpected alkene isomerization during its removal. These issues could be circumvented by employing a β-keto ester, which underwent smooth fluoride promoted decarboxylation without detectable isomerization. A triflate installed at an early stage served as a handle for the late-stage introduction of the northern side chain via a carbonylation/Brown allylation/cross-metathesis sequence. Further, we were able to access the closely related natural product 9-deacetoxy-14,15-deepoxyxeniculin and could show its base-mediated rearrangement to xeniafaraunol A. Our current work focuses on diversifying the northern side chain and skeletal editing of the 6/9-fused ring system to access related natural products.