Total Synthesis and Structural Assignment of (-)-Fusaequisin A.

(-)-Fusaequisin A is an irregularly assembled polyketide isolated from the ascomycete Fusarium equiseti. Fusaequisin A shares a carbon backbone with curvicollide C from the ascomycete Podospora curvicolla but its absolute configuration remained hitherto unsettled. Herein, we document the total synthesis of (-)-fusaequisin A and its 4-O-desmethyl derivative following a central-to-lateral building block strategy. Catalytic asymmetric Claisen rearrangement, Julia-Kocienski olefination and olefin cross-metathesis served as key C/C-connecting transformations. The constitution and absolute configuration of (-)-fusaequisin A was deduced and the original structural assignment was adjusted.

Total synthesis contributes to structural elucidation of natural products by enabling adjustment or settlement of the original structural assignment.[ , ] The isolation of the fungal metabolites curvicollide A−C (1 a–c) from an ascomycete of the genus Podospora (order Sordariales) was reported by Gloer et al. in 2004 (Figure 1). Structural elucidation was initially limited to the overall constitution and the relative configuration of the defining central γ‐lactone segment of the polyketide. We completed the structural assignment of (−)‐curvicollide C (1 c) by total synthesis of a small collection of diastereomers of (+)‐curvicollide C (2). During our quest for structural assignment of curvicollide C by total synthesis, Shiono et al. revealed the isolation of (−)‐fusaequisin A (3) from an ascomycete of the genus Fusarium (order Hypocreales) in 2013.[ , ] The structural assignment of fusaequisin A originally included the overall constitution and the relative configuration of the γ‐lactol methyl ether segment.

Figure 1

Curvicollide‐fusaequisin‐type polyketides.

We were occupied by the proposal that (−)‐curvicollide C from P. curvicolla (Sordariales) and (−)‐fusaequisin A from F. equiseti (Hypocreales) feature identical carbon scaffolds because the producing organisms are members of different taxonomical orders and were collected from different habitats in different regions of the world. Postulating identical absolute configurations for (−)‐curvicollide C and (−)‐fusaequisin A seemed tempting but ambitious at that time and required verification by total synthesis of (−)‐fusaequisin A (4). The free induction decays (FIDs) of the 1H and 13C NMR experiments with natural fusaequisin A were made available to us and were used for structural assignment.

Our synthetic planning was guided by a modular central‐to‐lateral synthetic design (Scheme 1). We sought to develop a unified approach wherein we could, in principle, access known, yet unknown, and derivatized members of the curvicollide‐fusaequisin family of polyketides. Retrosynthetic simplification of the proposed structure of (−)‐fusaequisin A yielded three chiral building blocks. The western C1−C5 aldol (−)‐5 and the eastern C15−C20 (+)‐6 sulfone were previously synthesized in our laboratory. The C8−C14 α,β‐enal 7 was designed to introduce the defining γ‐lactol ether segment. To allow for consecutive chain elongation in the eastern and the western direction, the termini of the central building block 7 were designed to be directly addressable. Accordingly, Julia‐Kocienski olefination between α,β‐enal 7 and sulfone (+)‐6 was planned with confidence. Projecting olefin cross‐metathesis (CM) to serve for the late‐stage construction of the C5−C8 diene moiety, however, was feared to be complicated or even derailed by reactivity and selectivity issues.

Scheme 1

Modular synthetic layout.

Our efforts toward 4 commenced with the enantioselective synthesis of the all‐trans configured α,β‐enal 7 starting with the catalytic asymmetric Claisen rearrangement of the Gosteli‐type allyl vinyl ether (E,Z)‐8 developed in our laboratory (Scheme 2). Subjecting (E,Z)‐8 to the 2,2′‐Isopropylidenebis[(4S)‐4‐tert‐butyl‐2‐oxazoline]‐derived copper(II) catalyst complex (S,S)‐9 at room temperature delivered the α‐keto ester (9R,10R)‐10 as a single configurational isomer in 98 % yield (1.95 g isolated mass). The chair‐like transition state structure 16 accounts for the formation of (9R,10R)‐10; peak‐affinity transition‐state stabilization drives the turnover under catalytic conditions at ambient temperature. Subsequent reduction with lithium aluminum hydride followed by oxidative glycol (11) cleavage tailored the α‐keto ester 10 to the aldehyde 12 (91 %). Pairing the MgBr2‐complexed aldehyde 12 with the in situ prepared vinyl lithium reagent 13 by nucleophilic addition delivered the allylic alcohol 14 as 2 : 1 mixture of diastereomers (91 %). In order to converge the C11 diastereomers, 14 was oxidized to the α,β‐enone 15 using the Dess‐Martin periodinane (DMP) in 93 % yield

Scheme 2

Catalytic enantioselective synthesis of α,β‐enone 15.

With the α,β‐enone 15 in hand, efforts to complete the asymmetric synthesis of the central building block 7 were launched (Scheme 3). Sodium bis(2‐methoxyethoxy)aluminum hydride (Red‐Al) reduction of 15 delivered the (11R)‐configured allylic alcohol 13 as a single diastereomer (76 %); the initial assignment of the absolute configuration of the non‐aldol stereotriad is in accordance with the Cram‐Felkin‐Anh model, and was later corroborated by NOE experiments (see below). Reductive removal of the benzyl protecting group delivered the diol 17. Subsequent efforts to directly promote γ‐lactolisation by regioselective oxidation at C9′ of 17 met with failure. Alternatively, Griffith‐Ley oxidation of 17 afforded γ‐lactone 18 in moderate yield (66 %). Reduced reaction times triggered formation of inseparable mixtures consisting of 18 and the corresponding γ‐lactole 19. Thus, we opted for a reliable redox sequence to access the cyclic hemiacetale 19; in this event, diisobutylaluminum hydride (DIBAL‐H) reduction of 18 provided 19 as an equimolar mixture of anomers (90 %). Subsequent diastereoselective γ‐lactol methyl ether formation was accomplished with methyl iodide in the presence of silver oxide at slightly elevated temperature (98 %); careful temperature control is required because mixtures of diastereomers were obtained at temperatures exceeding 35 °C. The results of a NOE experiment and coupling constants are in accordance with an all‐trans configuration of the γ‐lactone methyl ether 20. Removal of the allylic TBS ether and allylic oxidation proceeded uneventfully to deliver the central building block 7 in 83 % yield from 20.

Scheme 3

Synthesis of the central building block 7.

With the α,β‐enal 7 secured, efforts shifted next to the attachment of the eastern segment by Julia‐Kocienski olefination (Scheme 4). Lithiation of the chiral sulfone (+)‐6 was accomplished using an excess of lithium hexamethyldisilazide (LiHMDS) in THF; subsequent addition of the THF solution of the lithiated sulfone to a solution of the α,β‐enal 7 at −78 °C delivered the 12E,14E‐configured triene 22; rigorous temperature control is required to ensure the synthetically useful 14E/14Z‐diastereoselectivity. To forge ahead, the triene 22 was subjected to CM with methacrolein to install the trisubstituted E‐configured C7/C8 double bond. Initially feared as a potential obstacle due to steric hindrance and the presence of the 12,14‐diene moiety, the CM progressed slowly but steady under the conditions of Lipshutz to deliver the α,β‐enal 24 in useful yield (69 %), and some starting material (20 %). To set the stage for a second CM, 24 was subjected to Wittig methylenation to afford the tetraene 25 (88 %). Progressing on single digit milligram scale was considered sufficient for structural assignment. From key intermediate 25, 4‐O‐desmethyl fusaequisin A (−)‐27 was then completed by CM with the aldol (−)‐5. Surprisingly, however, comparison of the NMR data of synthetic (−)‐26 with those taken from the original FIDs for the natural product named (−)‐fusaequisin A matched perfectly, however, with one notable exception (Figure 2 and Supporting Information).

Scheme 4

Attaching the laterals to the center.

Figure 2

1H NMR spectra at 400 MHz in pyridine‐d5 (referenced to most downfield shifted signal of pyridine at δ=8.74 ppm); top: natural product 3 (FID provided by Y. Shiono); middle: synthetic (−)‐26; bottom: synthetic (−)‐4.

Conspicuously, a singlet at 3.62 ppm of the 1H NMR spectrum at 400 MHz pyridine‐D5 of the natural product (Figure 2, top) finds no counterpart in the 1H NMR spectrum of our synthetic 4‐O‐desmethyl fusaequisin A (−)‐26 (Figure 2, middle). However, when adding 1 equiv. of methanol to the NMR sample of synthetic (−)‐26, we observed an additional singlet at 3.62 ppm of an otherwise unchanged NMR spectrum of (−)‐26, indicating a possible contamination of the NMR sample of the natural product with methanol.

Perplexed by the unexpected congruence of the NMR spectra of (−)‐26 and the natural product, we concluded the total synthesis of (−)‐4 by 4‐O‐methylation of (−)‐26. As expected, the NMR spectra of synthetic (−)‐4 and those available for the natural product named (−)‐fusaequisin A deviate globally (Figure 2 and Supporting Information). Accordingly, reassignment of the reported constitution of fusaequisin A is inevitable. We found a convincing match between the NMR data of (−)‐4‐O‐desmethyl fusaequisin A (26) and the NMR data of the natural product, as processed from Shiono's FIDs (Figure 2). Hence, we suggest assigning the constitution and relative configuration of 26 to the natural product named fusaequisin A. The absolute configuration of (−)‐fusaequisin A (26) was assigned by the polarimetrically determined sense of optical rotation.

Reassigned fusaequisin A (26) from Fusarium equiseti embodies the γ‐lactol methyl ether of curvicollide C (2) from Podospora curvicolla (Scheme 4). Accordingly, fusaequisin A (26) should be renamed to fusaequisin C. We suggest the γ‐lactol methyl ethers of curvicollide A (“fusaequisin A”) and B (“fusaequisin B”) as natural products that await future isolation from Fusarium sp., Podospora sp., or any ascomycete equipped with the required biosynthetic machinery.

In conclusion, we have accomplished the first enantioselective synthesis of (−)‐4‐O‐desmethyl fusaequisin A (26) and of (−)‐fusaequisin A (4). A convergent central‐to‐lateral synthetic design was implemented. Our catalytic asymmetric Claisen rearrangement was utilized for building block synthesis. Olefin cross‐metathesis enabled rapid assembling of the C5−C8 2‐methyl‐1,3‐diene segment. Physical data evaluation provides sufficient evidence to justify an adjustment of the originally assigned constitution of the natural product named fusaequisin A (4). On grounds of NMR data similarity and polarimetry, we appoint (−)‐4‐O‐desmethyl fusaequisin A (26) as the natural product originally isolated from Fusarium equiseti by Shiono et al.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.