Fusarisetin A: Scalable Total Synthesis and Related Studies

Abstract

Fusarisetin A (1) is a recently isolated natural product that displays an unprecedented chemical motif and remarkable bioactivities as a potent cancer migration inhibitor. We describe here our studies leading to an efficient and scalable total synthesis of 1. Essential to the strategy was the development of a new route for the formation of a trans-decalin moiety of this compound and the application of an oxidative radical cyclization (ORC) reaction that produces fusarisetin A (1) from equisetin (2) via a bio-inspired process. TEMPO-induced and metal/O₂-promoted ORC reactions were evaluated. Biological screening in vitro confirms the reported potency of (+)-1. Importantly, ex vivo studies show that this compound is able to inhibit different types of cell migration. Moreover, the C₅ epimer of (+)-1 was also identified as a potent cancer migration inhibitor, while (−)-1 and 2 were found to be significantly less potent. The optimized synthesis is applicable on gram scale and provides a solid platform for analogue synthesis and methodical biological study.

Introduction

Despite the tremendous advances, cancer still represents an enormous medical challenge since, only in America, it is responsible for more than half a millions of deaths per year.¹ About 90% of these deaths are attributed to cancer metastasis, which is the ability of tumor cells to migrate from their tissue of origin and colonize elsewhere in the body.² When cancer is detected at a premetastatic stage, it can often be treated successfully either by local therapy (surgery, radiation) or by systemic therapy (chemotherapy, targeted therapy, hormonal therapy).³ However, when it is detected after it has metastasized, such treatments are much less successful. Along these lines, metastasis is considered as the “last frontier” in cancer management for which, to-date, there is no effective treatment.^4,5

In principle, it is possible to halt (or retard) cancer metastasis with the help of small molecules that inhibit cell migration.⁶ Recent advances in high-throughput screening and high content imaging techniques permit the identification of new cancer metastasis inhibitors from libraries of natural products or small molecules.^7,8 On the other hand, the availability of complex natural products via improved isolation techniques and streamlined synthetic strategies (or genetic engineering) allows evaluation of their effect in cell migration assays. Sceptrin⁹ and migrastatin¹⁰ represent a few recent examples of such efforts. Importantly, the development of scalable syntheses of these compounds^11,12 have resulted in the development of new chemical tools for the study of proteins involved in cancer metastasis and the discovery of potent cell migration inhibitors for further preclinical studies.^11b,12c,d

Recent efforts to identify potent inhibitors of cancer metastasis have led to the isolation of fusarisetin A (1) from a Fusarium species (Figure 1).¹³ This compound was found to inhibit cancer metastasis in MDA-MB-231 cells, a particularly aggressive breast cancer cell line. Specifically, 1 was found to inhibit acinar morphogenesis (IC₅₀ ca 77 µM), cell migration (IC₅₀ ca 7.7 µM) and cell invasion (IC₅₀ ca 26 µM) in these cell lines without any significant cytotoxicity in concentrations up to 77 µM. Interestingly, the proteomic profiling of 1 was found to be significantly different to those of other reference compounds. Moreover, 1 did not inhibit the phosphorylation of ERK1/2, AKT, c-Jun and p38 kinases in response to EGF treatment, as it is commonly observed with compounds that inhibit cancer metastasis by altering protein kinases.⁶ These findings suggest that the molecular target of fusarisetin A is different from those of known compounds and thus, its identification could produce new fundamental knowledge in the signal transduction pathways related to cancer metastasis. From a chemistry standpoint, fusarisetin A possesses an unprecedented pentacyclic ring system of which the CDE rings contain various polar functionalities (ketone, lactam, hemiketal and primary alcohol). The combination of impressive chemical structure and potent bioactivity is evidenced by the three publications describing various syntheses of 1 in about one year after its structure became known in the literature.^14,15,16 Intrigued by this molecule, we devised a project focused on the development of a scalable chemical synthesis that may allow the study of its chemical biology. Herein we report a detailed account on the synthesis of fusarisetin A. The developed strategy is short, efficient and stereoselective and is highlighted by the use of a key oxidative radical cyclization (ORC) reaction that allows conversion of equisetin (2) to fusarisetin A (1) following a likely biosynthetic pathway. Moreover, our biological studies confirm the reported bioactivity of 1 in vitro and indicate that this activity is maintained in whole tissue using ex vivo assays.

Fig. 1.

Retrosynthetic analysis of fusarisetin A (1) via equisetin (2) featuring a bioinspired oxidative radical cyclization (ORC) reaction (ROS: reactive oxygen species).

Results and discussion

Retrosynthetic analysis and strategic bond disconnections

Close inspection of the fusarisetin framework reveals the fusion of a trans-decalin unit (AB ring system) with a tetramic acid moiety (E ring). These rings can also be found in the structure of equisetin (2),¹⁷ another secondary metabolite produced by a Fusarium species, suggesting that both molecules may arise from a common biosynthetic pathway (Figure 1).¹⁸ Along these lines, we hypothesized that 1 derives biogenetically from oxidation of 2 upon exposure to reactive oxygen species (ROS).¹⁹ This biosynthetic scenario could account for the formation of stabilized radical 3 that, upon cyclization at the pendant alkene followed by trapping by ROS and hemiketalization, would produce 1. Translating this proposal to a synthetic plan, we envisioned that equisetin could undergo a 5-exo-trig oxidative radical cyclization (ORC)²⁰ to form the C₁–C₆ bond. The resulting C₅ radical could then be trapped by oxygen or a related ROS. Further bond disconnection of the equisetin motif suggested that 2 could be produced from β-ketoester 4, the trans-decalin motif of which (AB ring system) could be made via an intramolecular Diels-Alder reaction (IMDA).²¹ This scenario led to consider polyene 5 as the potential precursor of the IMDA. In turn, 5 could be synthesized from commercially available citronellal (6) whose motif contains the C₂₁ methyl group with the desired stereochemistry.²²

Synthesis of the trans-decalin moiety and evaluation of the TEMPO-induced ORC

The retrosynthetic plan shown in Figure 1 calls for departure of the synthesis from (R)-(+)-citronellal (6). Nonetheless, guided by the original assignment of fusarisetin A,¹³ we started our synthesis with commercially available (S)-(−)-citronellal (ent-6) (Scheme 1).²³ Inspiration for our studies came from previously reported syntheses of equisetin by the Danishefsky, Dixon and Shishido groups.²⁴ With an eye toward step-economy,²⁵ we sought to develop an alternative synthesis of this compound. To this end, ent-7 was synthesized from ent-6 via cross-metathesis with methacrolein using a Ru-carbene catalyst (Grubbs 2^nd generation, 5 mol%)²⁶ (75% yield). Alternatively, allylic oxidation of ent-6 with SeO₂/IBX can also form ent-7 in 65% yield. Although both reactions are scalable, in our subsequent studies we used the SeO₂/IBX method due to its lower cost. The two chemically differentiable carbonyl groups of this compound provide the possibility to install the polyene motif in a regioselective manner. Initial olefination studies of ent-7 under HWE or Julia conditions proved to be unsatisfactory.^24b–c,27 However, slow addition of the Wittig ylide, generated upon deprotonation of phosphonium salt 8,²⁸ to ent-7 afforded polyene ent-5 in 62% overall yield as a mixture of E/Z isomers (E: Z = ca 3:2). Photo-induced isomerisation of this mixture with catalytic amount of iodine²⁹ produced exclusively the trans polyene. Without purification, this compound was subjected to a Et₂AlCl-promoted IMDA reaction, that stereoselectively produced the desired trans-decalin aldehyde ent-9 (dr > 10:1, 82% yield). It is noted that the rapid construction of this trans-decalin motif could grant access to other biologically interesting natural products.³⁰ Treatment of ent-9 with ethyl bromoacetate under Reformatsky conditions followed by IBX oxidation yielded β-keto ester ent-4 in 91% combined yield. Aminolysis of this compound with serine derivative ent-10, followed by deprotection of the TBS group, produced ent-11 (43% yield overall). Dieckmann condensation of ent-11 produced a mixture of (+)-equisetin (ent-2) together with its C₃-epimer (ent-C₃-epi-2) (100% yield, dr = 1:1).²² It should be noted that the tendency of equisetin to epimerize at the C₃ center under basic conditions has been previously reported³¹ and has been observed consistently in our studies. Low temperature ¹H NMR experiments have also confirmed that equisetin exists exclusively in the enol form and not as a tautomeric mixture as recently reported.¹⁶ The structures of these compounds were also confirmed by comparison with the known data.^16,17,24,31

Scheme 1.

Synthesis of the trans-decalin moiety and study of the TEMPO-induced ORC. Reagents and conditions: (a) methacrolein (2.0 equiv), Grubbs^2nd gen. catalyst (5 mol%), CH₂Cl₂, 50 °C, 24 h, 75% (90% brsm); (b) SeO₂ (3 mol%), ^tBuOOH (4.0 equiv), salicylic acid (0.1 equiv), CH₂Cl₂, 96 h, then IBX (1.4 equiv), DMSO, 1.5 h, 65%; (c) 8 (1.0 equiv), n-BuLi (1.0 equiv), THF, −60 °C, 1 h, then −78 °C, ent-7, (see SI), 62%; (d) I₂ (5 mol%), sunlamp (visible light), CH₂Cl₂, 5 min, then −78 °C, Et₂AlCl (1.0 equiv), 24 h, 82%; (e) activated zinc dust (3.0 equiv), ethyl bromoacetate (1.2 equiv), PhH, 45 min, 90 °C; (f) IBX (2.0 equiv), DMSO, 80 °C, 10 min, 91% for 2 steps; (g) 10 (3.0 equiv), PhMe, reflux, 12 h; (h) HF (48% in H₂O, 10 equiv), MeCN, 15 min, 43% over 2 steps; (i) NaOMe (5.0 equiv), MeOH, 10 min, 100 % combined yield (dr = ca 1:1); (j) TBSCl (1.5 equiv), imidazole (2.0 equiv), CH₂Cl₂, 12 h, 90%; (k) 13 or Mn(OAc)₃ (2.0 equiv), TEMPO (3.0 equiv), DMF, microwave, 100 °C, 35%; (l) activated zinc dust (20 equiv), AcOH/THF/H₂O (3:1:1), 80 °C, 3 h, 30%.

With ent-2 in hand, we sought to explore ORC processes for the formation of the C ring of fusarisetin. It is worth noting that although radical reactions have often been used in natural products synthesis for the construction of C-C bonds,³² their application to the formation of C-O bonds remains limited.³³ A report by Jahn et al on the construction of 5-membered rings, using 1,3-dicarbonyl groups and alkenes under TEMPO conditions, provided a possible way for the desired transformation.³⁴ However, our initial studies with ent-2 gave unsatisfactory results, presumably due to the sensitivity of its C₃ hydroxymethyl group. To overcome this issue, we protected equisetin as its TBS ether ent-12 (90% yield). Gratifyingly this compound underwent the desired ORC, using ferrocenium hexafluorophosphate (13) or Mn(III) acetate as the oxidants,³⁵ to afford cyclized TEMPO-product ent-14 albeit in moderate yield (35%). Mechanistically, this reaction proceeds via a heat-promoted homolytic cleavage of the TEMPO-C₁ bond.³⁶ The resulting stabilized radical at C₁ reacts with the pendant C₅–C₆ alkene to generate the C₅-radical that can subsequently be trapped by the available TEMPO.³⁷ Reduction of the alkoxylamine bond of ent-14 under Zn/AcOH conditions³⁸ liberated the C₅-alcohol that underwent the desired hemiketalization, along with concomitant deprotection of the TBS group, to form a compound that was spectroscopically identified as the C₅-epimer of (−)-fusarisetin A (ent-C₅-(epi-1).¹⁴

The results of this study allowed us to draw several conclusions related to the TEMPO-mediated ORC reaction. As predicted, the TEMPO can indeed act as an ROS synthetic alternative and could form the C ring of ent-14 albeit in low yield. Gratifyingly, the stereochemistry of the C₁–C₆ bond was efficiently cotrolled by the structure of the decalin ring. Unfortunately, the stereochemistry of the C₅ center was not the desired one. Moreover, we encountered difficulties applying this reaction to a non-protected equisetin (ent-2). These considerations prompted us to apply the TEMPO-mediated ORC on a less functionalized substrate.

Total synthesis of (−)-fusarisetin A (ent-1)

β-Keto ester ent-4 appeared to be an attractive substrate for the TEMPO-mediated ORC, since it is less functionalized than equisetin and also contains an easily oxidizable C₁ center. With this in mind, ent-4 was treated with LiHMDS and the resulting C₁ enolate was in situ oxidized with 13 to afford, after quenching of the C₁ radical with TEMPO, compound ent-15. As expected, under these conditions (5 min, 0 °C) the ORC did not occur and ent-15 was isolated and fully characterized as a mixture of C₁-isomers (ca 2.5:1) in 99% yield. Heating this isomeric mixture at 90 °C over a period of 36 h gave rise to the tricyclic motif of ent-16 via the desired 5-exo-trig cyclization. Similarly with the above study, the formation of the C₁–C₆ bond proceeded with excellent stereocontrol, presumably due to the stereochemical bias of the decalin motif. Interestingly however, in this case we obtained a mixture of stereoisomers at C₅ (ca 1:1). It is worth mentioning that attempts to decrease the reaction time by raising the temperature proved to be problematic since they led to significant amounts of decarboxylated product ent-17. To further enhance the overall efficiency, we also examined the one-pot ORC and aminolysis sequence in presence of serine derivative ent-18. To our delight, this one-pot reaction gave rise to compound ent-19 (C₅ dr = ca 1:1) in 70% overall yield. To avoid the difficult separation of these diastereomers, ent-19 was directly treated with m-CPBA³⁹ to oxidatively cleave the N-O bond producing ent-20 in 95% yield. Regio- and stereo-selective reduction of this compound under Luche conditions^40,14 followed by a one-pot Dieckmann condensation/hemiketalization yielded (−)-fusarisetin A (ent-1) together with its C₅-epimer (dr = ca 4:1, 42% over 2 steps). Synthetic (−)-fusarisetin A was identical in all aspects with naturally occurring fusarisetin A (¹H-NMR, ¹³C-NMR and HR-MS), except for the optical rotation (synthetic: [α]_D²³ = −86.2 (c = 0.065 in MeOH); natural: [α]_D²⁵ = +84.6 (c = 0.2 in MeOH)¹³, reported synthetic (−)-1: [α]_D²⁷ = −88.0 (c = 0.15 in MeOH).¹⁴ The structure of ent-C₅-epi-1 was confirmed by comparison to the literature data.^14,16

Model studies on the metal-promoted ORC

Despite the overall efficiency of the one-pot TEMPO-mediated ORC/aminolysis reaction, the scale-up process suffers from significant decarboxylation of β-ketoester ent-16. Faced with this challenge, we evaluated alternative metal-promoted ORC reactions. To this end, we synthesized model system 24 that contains all key carbons needed for the proposed cyclization (Scheme 3). This compound was prepared via a sequence of 3 steps that included: (a) kinetic alkylation⁴¹ of ethyl acetoacetate (21) with crotyl bromide to form 22; (b) aminolysis of the ester group with N-methyl glycine methyl ester to produce 23; and (c) Dieckmann condensation (formation of the C₁–C₄ bond) to yield 24 (45% yield overall). Tetramic acid 24 was then subjected to various reagents and conditions in order to perform the desired ORC (Table 1). Scarce literature reports³³ indicate the feasibility of this transformation that, nonetheless, has never been applied to intramolecular systems or any natural product synthesis. It is known that certain high oxidation state metals, such as Mn(III), Co(II), Ag(II), Pd(II), Pb(IV) and Ce(III), can promote the addition of carbon radicals derived from ketones to alkenes.^20,42 On the other hand, molecular oxygen exists as a persistent triplet diradical in its ground state and as such it can react rapidly with carbon-centered radicals.⁴³ With this in mind, we treated 24 with various metals in the presence of O₂ and, in certain cases, were able to isolate peroxyhemiketal 25 (as C₅ isomers). Reduction of the peroxide motif of 25 then produced 26 (as C₅ isomers) representing the tricyclic core of fusarisetin A.

Scheme 3.

Model studies of metal-promoted ORC reactions. Reagents and conditions: (a) NaH (1.1 equiv), THF, 0 °C, 10 min; then n-BuLi (1.05 equiv), 0 °C, 10 min; crotyl bromide (1.07 equiv), 0 °C to RT, 2 h, 92%; (b) N-methyl glycine methyl ester•HCl (2.0 equiv), DMAP (2.0 equiv), Et₃N (3.0 equiv), PhMe, 110 °C, 12 h, 50%; (c) NaOMe (4.0 equiv), MeOH, 2 h, 97%.

[Table 1].

Selected conditions for the conversion of 24 to 25 and 26 via a metal-promoted ORC reaction.

Entry	Oxidant	Temp. °C)	Time	Yield of 25	Reduct- ant	Yield of 26
1	Co(OAc)2	70	5 min	20%	CuCl	80%
2	Co(OAc)2	70	5 min	20%	thiourea	n.r.[c]
3	Co(OAc)2	25	4 h	10%	CuCl	79%
4	Co(OAc)2[d]	25	12 h	n.r.	-	-
5	CoCl2	25	12 h	n.r.	-	-
6	Mn(OAc)3	25	12 h	5%	-	-
7	CeCl3	25	12 h	trace	-	-
8	Fe(III)[e]	25	12 h	15%	CuCl	81%
9	CAN	25	3 h	57%	CuCl	82%
10	CAN	–20	18 h	57%	CuCl	79%
11	CAN	70	5 min	20%	-	-
12	CAN[f]	25	3 h	40%	CuCl	80%
13	CAN[g]	70	3 h	30%	CuCl	81%

^[a,b][a] For a detailed screening study see SI;

^[b]all the reactions were performed under 1 bar of oxygen and 1 equiv of the oxidant in AcOH unless otherwise noted; Yields of 25 and 26 refer to isolated products;

^[c]no reaction occurred;

^[d]iPrOH was used as the solvent;

^[e]ferrocenium hexafluorophosphate;

^[f]0.1 equiv of CAN was used;

^[g]reaction opened to air. CAN = cerium(IV) ammonium nitrate.

The nature of the metal oxidant, oxygen pressure and reaction solvent/time were screened in order to optimize this ORC. Initial encouraging results were obtained using cobalt(II) acetate/AcOH/O₂ (1 bar) and afforded the desired cis-peroxyhemiketal 25 as a mixture of C₅ diastereomers (ratio 2:1) albeit in low yield (20%). Reduction of the cyclic peroxide group of 25 with CuCl yielded the corresponding tricyclic compounds 26a and 26b (ratio 2:1) in 80% overall yield (Table 1, entry 1).⁴⁴ Lower reaction temperature or replacement of the solvent/reductant led to unsatisfactory results (Table 1, entries 2–4). More than 20 metal or non-metal based oxidants were then further screened (see SI) under AcOH/O₂ conditions. Mn(III)-, Fe(III)- and Ce(III)-based oxidations were found to be inefficient for the conversion of 24 to 25 (Table 1, entries 6–8). Gratifyingly, cerium(IV) ammonium nitrate (CAN) significantly increased the yield of 25 to 57% (Table 1, entry 9). Decreasing the reaction temperature did not affect the diastereoselectivity at C₅ (Table 1, entry 10), while higher temperatures decreased the yield (Table 1, entry 11). Air or catalytic amounts of CAN (10 mol%) can also be used for the conversion of 24 to 25 but in this case the formation of 25 is low yielding (Table 1, entries 12–13). Nonetheless, in all cases the reaction proceeded in a highly stereocontrolled manner affording only two of eight possible diastereomers (ratio ca 1.5:1 ~ 2:1).⁴⁴ It is worth mentioning that, independently of this study, Gao et al recently reported the conversion of (−)-equisetin to (+)-fusarisetin A using Mn(III)/O₂ conditions.¹⁶

Total synthesis of (−)-equisetin (2) and (+)-fusarisetin A (1)

Encouraged with the above results, we proceeded to implement the optimized ORC conditions to the synthesis of (+)-fusarisetin A using (−)-equisetin as the key synthetic intermediate (Scheme 4). Aldehyde 9 was rapidly and stereoselectively constructed from (R)-citronellal (6) in decagram-scale (35% over 3 steps) following the above procedure. Conversion of 9 to 4 proceeded under Reformatsky conditions followed by oxidation of the resulting alcohol with Dess-Martin periodinane (2 steps, 92% overall yield, ca 7 grams prepared). It is worth noting that at this scale the IBX oxidation produced significant amounts of decarboxylated material. Aminolysis of ester 4 under various conditions, such as DMAP^14,15 and NHC-based reagents,⁴⁵ with N-methyl serine methylester (18)⁴⁶ afforded 11 in low yield. However, mild hydrolysis of 4 with ethanolic KOH (rt, 96 h) quantitatively produced the corresponding carboxylic acid that, upon coupling with 18 under HATU conditions afforded the desired amide 11 in 90% yield. Dieckmann condensation of 11 quantitatively produced a mixture of (−)-equisetin (2) and C₃-epi-equisetin (C₃-epi-2) (100%, dr = ca 1:1).^16,17,24,31

Scheme 4.

Scalable total synthesis of (−)-equisetin (2) and (+)-fusarisetin A (1). Reagents and conditions: (a) activated zinc dust (5.0 equiv), ethyl bromoacetate (3.0 equiv), PhH, 90 °C, 45 min; (b) Dess-Martin periodinane (2.0 equiv), CH₂Cl₂, RT, 2 h, 92% for 2 steps; (c) KOH (5.6 equiv), EtOH, RT, 96 h, then 18 (1.2 equiv), HATU (1.1 equiv), CH₂Cl₂, DMF, 0 °C, iPr₂NEt (3.0 equiv), then RT, 2 h, 90%; (d) NaOMe (5.0 equiv), MeOH, RT, 10 min, 100% combined yield (dr = 1:1); (e) CAN (1.0 equiv), AcOH, O₂, RT, 3 h, then silica pad, thiourea (10 equiv), MeOH, 70 °C, 1 h, 62% combined yield (dr = 1.3:1).

Gratifyingly, the previously defined ORC conditions were successfully applied for the conversion of (−)-equisetin (2) to (+)-fusarisetin A (1). Specifically, oxidation of 2 under CAN/AcOH/O₂ conditions produced an inseparable mixture of peroxy-fusarisetin A (27) and its C₅ epimer (C₅-epi-27) (dr = 1.3:1). This mixture was further reduced with thiourea (10 equiv) to afford (+)-fusarisetin A (1) together with its C₅ epimer (C₅-epi-1) (62% overall, dr = 1.3:1).⁴⁷ The structures of both C₅ epimers of 27 and 1 have been confirmed by ¹H-NMR, ¹³C NMR and HR-MS analysis.^13–16 As observed previously, the stereoselectivity of this ORC reaction is substrate-controlled and affords the desired stereochemistry at the C₁ and C₆ centers.

We were also able to use the mixture of equisetin (2) and C₃-epi-equisetin (C₃-epi-2) in the ORC reaction. In fact, 1.1 grams of this mixture (dr = 1:1) were treated under CAN/AcOH/O₂ conditions and the resulting crude mixture of peroxy-fusarisetins was reduced with excess thiourea. Purification of this mixture produced 200 mg of (+)-fusarisetin A (1). Notably, the whole synthetic process from decalin 9 to (+)-fusarisetin A was performed on gram-scale and requires only one purification via column chromatography. In summary, the syntheses of both 1 and 2 are scalable, redox-/step-economic and protecting-group free.⁴⁸

Biological studies of 1, 2 and related structures

To confirm and expand upon the previously reported findings, we evaluated the biological activity of (+)-fusarisetin A (1) in a scratch-wound assay and in a Boyden Chamber Transwell assay. The first assay (Figure 2) involves inflicting a scratch wound in a confluent cell monolayer and measuring the migration of cells.⁴⁹ We were pleased to find that synthetic 1 inhibited migration of these cells at concentrations as low as 1 µg/mL (Figure 2C) as compared to vehicle control (Figure 2B). Importantly, removing 1 from the cells followed by incubation with fresh growth media allowed cells to migrate in a similar fashion to the control experiment (Figure 2D). This observation demonstrates that the effect of 1 on these cells is reversible, in turn suggesting that (+)-fusarisetin A has little to no cytotoxicity at 1 µg/mL concentration.

Fig. 2.

(+)-Fusarisetin A (1) reversibly inhibits the migration of MDA-MB-231 breast cancer cells in an in vitro scratch-wound assay. Cells grown as a confluent monolayer were scratched, photographed, and incubated with DMSO (control) and 1 (1 µg/mL in DMSO). Micrographs (10× magnification) are presented showing cell migration across the scratch. (A): wound created at the time of the scratch (time = 0 h); (B) cell migration in presence of DMSO (time= 48 h); (C) cell migration in presence of 1 (1 µg/mL), (time = 48 h); (D) wounded cells in panel C were rinsed with PBS to remove 1 and then incubated in fresh growth media for an additional 24 h.

We then performed Transwell migration assays using increasing concentrations of 1 normalized with the appropriate DMSO controls. This assay measures the capacity of cells to migrate across a porous membrane using serum-rich media as a chemo-attractant.⁵⁰ As seen in Figure 3, cell migration was significantly inhibited at 3.0 and 6.0 µg/mL, while almost complete inhibition is observed at 12.0 µg/mL. Even at this concentration we did not observe any changes in the cell morphology, suggesting that 1 exhibits low cytotoxicity. The results of these two assays confirm the reported biological activity of fusarisetin A in vitro.

Fig. 3.

(+)-Fusarisetin A (1) inhibits the migration of MDA-MB-231 breast cancer cells in a Transwell migration assay. Cells were plated in a Boyden chamber with 1 or vehicle control (DMSO) for 24 h at the indicated concentrations. Amount of migrated cells is presented relative to the corresponding DMSO control.

Encouraged by these findings, we then evaluated (+)-fusarisetin A in an ex vivo assay measuring migration of cells from a 5 mm mouse skin biopsy (Figure 4).⁵¹ We observed that both keratinocyte and fibroblast migration is inhibited upon exposure to 1 (10 µg/mL). Specifically, a substantial amount of cell migration is observed 5 days after plating the skin explant (Figure 4B), as compared to the initial time of plating (Figure 4A). In contrast, when explants were exposed to 10 µg/mL of 1 (Figure 4C), there was no detectable migration of keratinocytes from the explants, while fibroblast migration was reduced by approximately 80%. The observed ability of fusarisetin A to inhibit cell migration from skin explants is particularly exciting. The migration of fibroblasts is an example of mesenchymal cell migration whereas keratinocytes move via collective cell migration.⁵² Interestingly, compounds that target one type of migration have had disappointing results in clinical studies as the cancer cells are able to adapt and switch between different modes of migration.⁵³ Since fusarisetin A can significantly inhibit both types of migratory behavior, it could provide a powerful tool to circumvent the ability of cancer cells to alter their mode of motility if one pathway is inhibited.

Fig. 4.

(+)-Fusarisetin A (1) inhibits cell migration in an ex vivo mouse skin assay. Ker: keratinocyte migration (red line); fib: fibroblast migration (blue line). (A) 5mm explant of mouse skin in a tissue culture dish (time = 0 days); (B) Migration of cells from skin explant incubated for 5 days in growth media containing control amount of DMSO; (C) Migration of cells from skin explant incubated for 5 days in growth media containing 10 µg/mL of fusarisetin A. The blue dotted line in Figure 4B shows the migration area of keratinocytes. Solid lines indicate migration distance.

Having demonstrated the cell-migration inhibitory properties of 1 both in vitro and ex vivo we then screened selected compounds containing the fusarisetin framework using the scratch-wound assay.⁵⁴ We observed that natural (−)-equisetin (2) and its enantiomer ent-2 do not exhibit any activity at the concentrations tested (Figure 5). Similarly, ent-fusarisetin A (ent-1) was found to be inactive. However, the C₅ epimer of natural fusarisetin A (C₅-epi-1) was found to display similar activity to that of (+)-1. These initial findings attest to the importance of the CDE ring structure and suggest that only the naturally occurring enantiomer of fusarisetin A could be used as a motif for the identification of new inhibitors of cell migration.

Fig. 5.

Screening of fusarisetin A (1), equisetin (2) and selected synthetic intermediates in a scratch-wound assay at two different concentrations (7.7 and 15.4 µM). Values are given as relative migration in comparison to DMSO control.

Conclusions

We report here a concise, efficient, and protecting group-free synthesis of fusarisetin A (1). Key to our synthetic strategy is the implementation of a bioinspired oxidative radical cyclization (ORC) reaction that forms the C ring of 1 via stereoselective construction of the C₁–C₆ bond. Subsequent oxidation at the C₅ center allows formation of the D ring of 1 ultimately converting equisetin (2) to fusarisetin A (1). The TEMPO-mediated ORC reaction could be successfully applied for the conversion of β-ketoester 4 to a tricyclic motif 19 that, upon Dieckmann condensation/ hemiketalization, formed fusarisetin (1) together with its C₅ epimer (C₅-epi-1). However, treatment of equisetin under these conditions produced exclusively the C₅ epimer of fusarisetin A (C₅-epi-1). On the other hand, metal-mediated ORC reactions, such as Co(OAc)₂, Mn(OAc)₃, ferrocenium- and cerium(IV)-salts, could be successfully applied in a model system for the construction of the CDE ring of 1. Moreover, the conversion of 2 to 1 was best achieved using CAN/AcOH/O₂. Overall, the optimized synthesis of (+)-fusarisetin A proceeds in 8 steps and about 5% yield starting from commercially available R-(+)-citronellal. We have also confirmed that (+)-fusarisetin A exhibits potent inhibitory activities against cancer metastasis in vitro and demonstrated its capability to inhibit different types of cell migration in mice skin. Interestingly, equisetin (either enantiomer) and (−)-fusarisetin A were found to be inactive in these assays, while C₅-epi-1 displayed comparable activities to that of the natural product. In turn, this suggests that the motif of (+)-fusarisetin A could lead to new potent cancer metastasis inhibitors. The scalable synthetic strategy presented here could pave the way for more detailed structure-activity relationship and chemical biology studies.

Scheme 2.

First generation total synthesis of (−)-fusarisetin A (ent-1) via a TEMPO-mediated ORC. Reagents and conditions: (a) LiHMDS (1.5 equiv), 1,2-DME, −78 °C, 30 min, then 0 °C, TEMPO (1.05 equiv), Cp₂FePF₆ (13, 2.0 equiv), 5 min, 99%; (b) ent-18, DMAP (2.0 equiv), PhMe, 4Å MS, 90 °C, 36 h, 70% (d.r. = ca 1:1); (c) mCPBA (1.2 equiv), CH₂Cl₂, 0 °C, 15 min, 95%; (d) NaBH₄ (1.2 equiv), CeCl₃•7H₂O (1.5 equiv), MeOH, −20 °C (d.r. = ca 4:1); (e) NaOMe (5.0 equiv), MeOH, 10 min, 42% for 2 steps.