Total Synthesis of Tagetitoxin

Abstract

The intriguing structure of 1, a longstanding challenge in natural product synthesis, has been the subject of multiple revisions and has been confirmed through total synthesis. The route commences from a renewable furan starting material and features a number of unusual transformations (such as rearrangements, bromocyclization, P(V)-based phosphate installation) to arrive at the target in 15 steps. As the route was designed to enable access to both enantiomers, the absolute configuration of the natural product could be assigned using a bioassay on (+)-1 and (−)-1.

Graphical Abstract

For nearly a century, the phytotoxic alkaloid tagetitoxin (1) has piqued the interest of the multiple scientific communities on the basis of its bioactivity, structural ambiguity, and synthetic challenge. Tagetitoxin was first introduced to the world in 1937 as an enigmatic plant pathogen in the horticulture community,¹ until molecular biology revealed its then-unprecedented inhibitory effects on RNA Polymerase over fifty years later.² From a structural standpoint, the precise connectivity of 1 has been debated for nearly four decades (Figure 1). Mitchell’s inaugural structural assignment 1a,³ which centered on an unusual eight-membered thiocane heterocycle, was reevaluated again by the same group, and revised to an equally unusual set of possible isomeric [3.3.1]-bicycles (1b/1c).⁴ Despite several synthetic efforts towards this structure,⁵ the proposed connectivity was never confirmed, and the target proved elusive to synthesis. Recently, Aliev and co-workers reported a detailed NMR study on tagetitoxin that, with the aid of HMBC analysis, led to the disclosure of yet another proposed structure: the trans-6,5 bicycle 1.⁶ Neither the absolute configuration nor the optical rotation of natural 1 has been disclosed. The ominous synthetic challenge of 1 is clear as it harbors a fully-oxidized cyclopentane core, multiple polar functional groups, and contains more heteroatoms than carbon atoms. Intrigued both by the controversial structure and the unique bioactivity of 1, an approach was pursued that would access both enantiomers and shed light on its constitution. Facilitated by a variety of unusual transformations, this Communication delineates a successful total synthesis of both enantiomers of 1, enabling unambiguous assignment of the natural isolate thereby setting the stage for future biological explorations.

Figure 1.

Structural revisions of tagetitoxin (1) and retrosynthetic analysis for the current total synthesis.

Several salient structural features of 1 played a key role in drafting the retrosynthetic blueprint: (1) the fully-substituted cyclopentane core with unusually placed heteroatoms including the C7–S bond; (2) the trans-fused 1,4-oxathiane ring system; (3) potential concerns over retro-aldol reactivity at C5 and C8; (4) the C8-phosphate; (5) multiple charged species residing in close proximity (including 2 carboxylates, an amine, a phosphate); and (6) the documented instability of 1 to mildly acidic conditions.¹ The SI details many of the dead-ends and detours that provided key insights into the ultimately successful plan. Since the absolute configuration of the natural isolate was unknown, a racemic approach was pursued with a late-stage resolution. To facilitate the separation of enantiomers, generate a crystalline intermediate, and simultaneously install the phosphate group, it was hypothesized that the chiral Ψ-reagent⁷ would accomplish all three goals through separation of diastereomeric adducts 2 at a late stage, followed by hydrolysis. The C5/6 syn oxygenation was a keying element for dihydroxylation, and it was anticipated that subsequent selective cyclization would forge the trans-fused 1,4-oxathiane leading to cyclopentene 3. This disconnection led to a dehydrocysteine precursor, which, in turn could be accessed from thiol 4 and the appropriate amino acid. The requisite thiol was envisaged to arise from a thio-[3,3]-rearrangement⁸ of differentially substituted diol 5. An oxidative rearrangement of furfural-derived furans such as 6, reminiscent of the venerable Piancatelli reaction,⁹ seemed perfectly suited to access such a structure.

Our successful execution of this retrosynthetic vision leading to the synthesis of 1 is outlined in Scheme 1. Imine 7, derived from furfural (ca. $0.04 / gram) via condensation with tert-butylsulfonamide (gram-scale), was subjected to Mannich addition of the known TBS-based masked acyl cyanide (MAC) reagent.¹⁰ In the same pot, the resulting adduct was unmasked with TBAF in the presence of 2,2,2-trifluoroethanol to give amino ester 8. The choice of alcohol was dictated largely by product stability, as retro-Mannich products predominated when other nucleophiles where employed (see SI). This one-pot MAC-based approach proved to be readily scalable (multigram scale) and high-yielding to deliver the desired precursor for the ensuing oxidative rearrangement. To that end, subjecting 8 to singlet oxygen via photosensitized conditions, and quenching the resultant endo-peroxide sequentially with dimethylsulfide and SiO₂ at cryogenic temperature cleanly delivered enone 9. Notably, the desired anti-isomer was preferentially formed (5.4:1 dr), which is hypothesized to be the result of a rapidly-interconverting retro-aldol / aldol equilibrium prior to quenching (i.e., thermodynamic control). The choice of reductant also proved essential, as more nucleophilic reducing agents led to a range of side products (see SI for details). The formation of quaternary amino esters in such an oxidative Piancatelli-like rearrangement has not been previously disclosed.⁹ Preliminary studies using an Ellman-based sulfinimine¹¹ instead of a sulfonamide were promising but not pursued for the final route as the absolute configuration of 1 was unknown (see SI). Acylation of the secondary alcohol proceeded smoothly using p-cyanobenzoyl chloride in the presence of DMAP to deliver 10 on multigram scale (verified by X-ray crystallography).

Scheme 1.

^a For reagents and conditions, see the SI.

Having secured significant quantities of the requisite cyclopentenone core, we turned attention to the problem of stereoselective formation of the key C7–S bond. Numerous strategies were explored (see SI for details) before settling on the thio-[3,3] rearrangement. This was guided by the notion that the stereochemistry of an allylic alcohol such as 5 (Figure 1) could be transposed to control the C–S bearing stereocenter. It was anticipated that the stereoselectivity of the requisite 1,2-reduction of 10 could be controlled by the bulky benzoate group at C8 (i.e., kinetic control driven by sterics). Indeed, Luche reduction of 10 proceeded with high diastereoselectivity (relative stereochemistry confirmed by nOE), and after treatment with thio-CDI, afforded the desired rearrangement precursor 11. To our delight, simple heating of a dilute solution of 11 in toluene in the presence of a radical inhibitor, followed by acidic hydrolysis, delivered thiol 12 in 84% yield (gram-scale). To prepare for the annulation of the critical 1,4-oxathiane ring system, bromo-enamide 13 was appended in 65% yield furnishing 14 (gram-scale) via hetero-Michael addition/elimination. The N-Boc group of 13 was found to be essential as it not only increased electrophilicity of the π-system, but also greatly improved the selectivity of the subsequent C5/6-dihydroxylation. Accordingly, catalytic dihydroxylation¹² of the resulting olefin delivered diol 15 as a single diastereomer in 81% yield after acidic workup (with concomitant Boc deprotection), thereby furnishing the fully-functionalized core of 1. The structure of 15 was also confirmed by X-ray crystallography.

With cyclopentane 15 in hand, attention turned to the challenging task of forming the trans-fused 1,4-oxathiane hetereocycle.¹³ In addition to the strained nature of the ring junction, the prospect of C–O bond formation at such an electronically-ambiphilic position was a cause for concern. After extensive experimentation, it was discovered that dibromo-5,5-dimethylhydantoin could be successfully leveraged to effect bromoetherification and deliver heterocycle 16 in 40–45% yield (100 mg scale). Numerous carefully-defined reaction parameters such as concentration, temperature, stoichiometry, Br⁺ source, and reaction time were all essential in maximizing the yield of the desired product (see SI for details). The formation of a single diastereomer (as verified by X-ray crystallography) is notable and likely stems from an anomeric effect preferentially stabilizing the axial configuration of the bromide in the transition state.¹⁴ The precise mechanism of this bromocyclization is currently unclear; however, in situ NMR experiments indicate the initial formation of two observable intermediates that converge to 16 (consistent with a Curtin-Hammett scenario; see SI for details).

Given that all connectivity and atoms found in the natural product (with the exception of the C8 phosphate) were installed, only seemingly-simple operations (i.e., C–Br reduction, ester hydrolysis, deprotection, and phosphorylation) remained to complete the total synthesis of 1. In reality, however, successful development of the endgame strategy was the product of a disproportionate level of experimentation and strategic reevaluation (see SI). Of particular importance was the precise choreography of events, largely dictated by empirically-determined compound stability. Ultimately, the N-tert-butysulfonamide could be removed from saturated bicycle 16 via treatment with triflic acid in the presence of cation scavenger.¹⁵ Reductive radical debromination of the resultant amine 17, followed by hemiaminal formation in a single pot, furnished tricycle 18 (80% yield). Masking of this late-stage intermediate as the hemiaminal was not only required to achieve selectivity in the subsequent O-phosphorylation; this hemiaminal was also crucial in imparting stability and solubility in organic solvents, due to the ever-increasing polarity of intermediates leading to 1.

Two pivotal final manipulations remained: O-phosphorylation and ester hydrolysis. The recently disclosed set of chiral P(V) reagents (Ψ),⁷ capable of controlling the chirality of P-based linkages, appeared ideal for achieving the former task. Given the inconclusive nature surrounding the absolute configuration, the lack of authentic material,²⁰ and absence of optical rotation measurements of natural 1, we hypothesized that application of Ψ in this context might allow for phosphate installation, crystallization, and chiral resolution. To this end, selective methanolysis of the C8 benzoate left the carboxylic esters intact (essential for maintaining solubility in organic solvents) and set the stage for P-incorporation. Exposure of this alcohol to (+)-Ψ furnished diastereomers 19a and 19b, which were readily separable by SFC. The structure and absolute configuration of 19a was confirmed by X-ray crystallography. Oxidation of phosphorothioates 19a and 19b (P–S to P–O) using SeO₂,¹⁶ followed by global hydrolysis, independently delivered both (+)-1 and (−)-1. As depicted in Scheme 1, the final step consisted of no less than four independent hydrolysis events to arrive at 1. Thus, treatment of 19 with TMSOK concurrently deprotected the masked phosphate and methyl esters.¹⁷ Subsequent addition of a hydroxylammonium chloride buffer provided sufficient nucleophilicity to deprotect the hemiaminal, while its mild acidity obviated any decomposition pathways that would consume the final product.¹⁸ Notably, more standard conditions to effect these transformations (i.e., aqueous inorganic bases, formic acid) were met with inferior results, due to the notorious instability of the product (1). This final global hydrolysis event was successful in delivering the target compound, which matched all published spectral data from the most recent structural proposal. The highly charged nature of 1 required purification via ion exchange chromatography (DEAE resin, H₂O wash; then 0.2–0.5 M NH₄HCO₃); notably, the compound is not stable to HPLC purification. The critical role that Ψ played in this synthesis cannot be overstated and go beyond the three points mentioned above as the most commonly used methods for phosphate installation using both P(III)- or P(V)-based reagents failed due to either poor chemoselectivity or lack of reactivity, respectively (see SI for details).

Having obtained ample quantities of (+) and (−) 1, attention turned to elucidating the absolute configuration of tagetitoxin. Given the absence of optical rotation data from previous reports (and the lack of any natural sample),¹⁹ the only remaining tool at our disposal would be a bioactivity assay, with the hypothesis that only one enantiomer would be active. While such a tactic for absolute configuration determination is logical and likely has been used before, we are unaware of such precedent. To probe this hypothesis, a direct RNA polymerase inhibition assay was performed. Each enantiomer was assessed for its ability to inactivate E. coli RNA polymerase towards a DNA template as quantified by a double strand-selective intercalating fluorescent dye.²⁰ To this end, we were pleased to observe significant RNA polymerase inhibition in the presence of only (+)-1, while (−)-1 was totally devoid of this inhibitory activity (Figure 2). Thus, we conclude that (+)-1 is almost assuredly the natural isolate produced in Nature.

Figure 2.

Direct RNA polymerase inhibition assay revealed that (+)-1 is the enantiomer produced in Nature.

Nearly 80 years following its first detection in Nature, the enigmatic structure and absolute configuration of 1 has been conclusively settled. The challenge associated with a small, stereochemically ornate natural product bearing more heteroatoms (N, O, S, and P) than carbon atoms required unique considerations of chemoselectivity and oxidation choreography. This was achieved through the design and implementation of a total synthesis predicated on numerous unusual maneuvers to cement the densely functionalized, highly polar central cyclopentane core. Some key features of this design include the use of an inexpensive furan-based starting material, strategic rearrangements (Piancatelli-like, thio-[3,3]), a bromocyclization (to install the exotic trans-fused 1,4-oxathiane), and careful selection of orthogonal blocking groups. The inaugural use of Ψ reagents in total synthesis enabled phosphate installation (conventional reagents failed), diastereomer separation, and crystallization to assign absolute configuration. Finally, enlisting a bioassay for absolute configuration determination when natural supplies have been depleted and optical rotation measurements are absent is of note.