Bioinspired concise synthesis of caged Sesquiterpenoids Artatrovirenols A and B

Abstract

Artatrovirenols A and B are two newly isolated sesquiterpenoids with a complex caged framework. We report herein a concise synthesis of artatrovirenols A and B in 9 and 8 steps, respectively. The complex caged tetracycle is rapidly constructed from a known planar guaiane-type precursor through a bioinspired intramolecular [4 + 2] cyclization to firstly access artatrovirenol B, which is further transformed into artatrovirenol A through a biomimetic epoxidation-mediated lactonization reaction. This synthesis establishes a concise asymmetric approach to access artatrovirenols A and B, and also provides insightful evidence to their biogenetic pathway in nature.

Artatrovirenols A and B are two natural sesquiterpenoids with a complex tetracyclic backbone, and their chemical total synthesis is challenging. Here, the authors report a concise synthesis of artatrovirenols A and B in 9 and 8 steps, respectively, featuring a bioinspired [4 + 2] cycloaddition reaction to construct the tetracyclic core.

Introduction

Sesquiterpenoids are a large and evergrowing family of natural products with diversified skeletons and significant bioactivities^1,2. Although only 15 carbon atoms are involved, over 50000 members have been identified and categorized into more than 300 distinct carbon skeletons to date (Fig. 1a)^3–5. Among them, compared to the “planar” congeners like eudesmanes (1) and guaianes (2), “caged” sesquiterpenoids like longibornanes (3)^6,7 and longipinanes (4)^8–10 exhibit highly congested frameworks and functionalities, thereby, pose overwhelming challenges in a synthetic perspective. Recently, caged sesquiterpenoids serve as rich sources of inspirations of new synthetic strategies, culminating in elegant total syntheses in synthetic community from the research groups including Sarpong^11–14, Maimone^15–18, Rychnovsky¹⁹, Zhang^20,21, Yang²², and Kalesse²³. Our group recently disclosed a radical cyclization and semipinacol rearrangement strategy to achieve a collective total syntheses of longipinane-type sesquiterpenoids^24,25. Therefore, explorations on the syntheses of those newly isolated sesquiterpenoids with complex caged skeletons would unarguably motivate the inspiration of discovery of new chemistry and synthetic strategies.

Fig. 1. Structures of sesquiterpenoids.

a Typical sesquiterpenoid skeletons. b The structures of three new caged sesquiterpenoids.

In 2020, Chen and coworkers identified artatrovirenols A (5) and B (6) (Fig. 1b) from Artemisia atrovirens, a traditional Chinese herb used to treat stomach diseases and relieve pain²⁶. Biologically, artatrovirenol A showed cytotoxity against three human hepatoma cell lines and acted as a potential candidate for antihepatoma drug. Subsequently, a similar compound daphnenoid A (7) was isolated by Song and coworkers from Daphne penicillata²⁷. Structurally, these three sesquiterpenoids exhibit a caged tetracyclo[5.3.1.1.^4,110^1,5]dodecane framework (8). To the best of our knowledge, only these three natural sesquiterpenoids display such a complex caged network to date. Remarkably, high oxidation states in the forms of alkene, alcohol, ketone, and carboxylic acid and its derivatives (ester and lactone) at diverse positions are densely distributed in the system. Moreover, up to eight stereogenic centers are embedded in the framework, in a contiguous fashion, particularly, including three all-carbon quaternary centers located at the bridge-head positions (C1, C4 and C11). These structural features render them formidable target molecules in light of total synthesis.

Biosynthetically, the biogenetic pathway of the carbon skeleton of these sesquiterpenoids is currently not determined and lacks evidence of chemical transformations^26–28. As proposed by Chen and coworkers, the symbiotic planar guaiane-type sesquiterpenoid arglabin (9) might be the precursor in nature to generate a cyclopentadiene moiety in 10 to react with the acrylic acid side chain through an intramolecular Diels–Alder (IMDA) cycloaddition process, providing the norbornene core of the key framework 11 (Fig. 2a)²⁶. Dehydration of 11 would afford artatrovirenol B (6), which would further undergo a site- and stereo-selective epoxidation and lactonization process to deliver artatrovirenol A (5). As for the biogenetic pathway of daphnenoid A (7), Song also proposed that an intramolecular formal [4 + 2] cyclization in a stepwise fashion might convert the planar guaiane-type precursor into this caged framework²⁷. These proposals might be reasonable as the the energetic feasibility of each of these key cyclization reactions was assessed by D. J. Tantillo using density functional theory calculations²⁸, but it lacks evidence of chemical transformations to probe its possibility. From the synthetic perspective, the IMDA reaction is challenging to occur for the concerns of: 1) the free cyclopentadiene is prone to be deprotonated to form an aromatic cyclopentadienyl anion, where the olefin isomerization within the cycle would affect the IMDA site selectivity, 2) the cyclic diene and exocyclic dienophile are too far to react in an intramolecular fashion, and 3) the tetracyclic scaffold of 11 is highly strained and sterically disfavored to be formed.

Fig. 2. The possible biogenetic pathway of artatrovirenols and Zhu’s total synthesis of artatrovirenol A.

a The proposed biogenetic pathway. b Zhu’s de novo synthesis of artatrovirenol A.

With the above-mentioned challenges in the proposed biogenetic pathway, total syntheses of these three intriguing caged sesquiterpenoids through a bioinspired approach seemly serve as a formidable mission, which prompted Zhu and coworkers²⁹ to adopt a de novo approach for total synthesis of artatrovirenol A (5), involving an intermolecular Diels–Alder reaction of dienophile 12 with isoprene to access bicyclic intermediate 13 followed by a Mukaiyama-Michael addition to chiral cyclopentenone 14 and a key de Mayo reaction to ultimately reach the final framework (Fig. 2b). The elegant strategy designed and the new chemistry discovered in Zhu’s work demonstrated the power of mankind in sophisticated editing of molecular structures and the synthesis provided a classic example in modern chemical synthesis. This de novo approach inevitably contains 18 steps of chemical transformations from the known precursor 12 (made in 5 steps from D-mannitol). Driven by the complexity of the tetracyclic skeleton of 5–7 and in connection with our interest in the synthesis of the challenging caged natural products^24,25,30, we initiated the project to explore the probability of the aforementioned proposed biogenetic pathway and achieve a bioinspired synthetic approach to access both artatrovirenols A (5) and B (6). Notably, when our work was in the final stage, Robertson and coworkers³¹ reported a seminal work on construction of the key framework through a concerted thermo-induced IMDA reaction, which unfortunately failed to reach the final natural products. Herein, we present our syntheses of artatrovirenols A (5) and B (6) based on a bioinspired intramolecular formal [4 + 2] cycloaddition of cyclopentenone and acrylate moieties and a biomimetic epoxidation-mediated lactonization strategy, culminating in a concise route of 5 and 6 in 9 and 8 steps, respectively, from a readily accessible known guaiane-type precursor.

Results

Retrosynthetic design

Retrosynthetically, the C9-hydroxyl γ-lactone moiety of artatrovirenol A (5) was traced back to the endocyclic olefin and carboxylic acid functionalities of artatrovirenol B (6) through a biomimetic epoxidation and lactonization process (Fig. 3). Then, the key framework of 6 was supposed to be constructed through a formal intramolecular [4 + 2] cycloaddition reaction. To inhibit the unwanted isomerization of free cyclopentadiene, we designed the enolate 15 as the putative reactive [4 + 2] cycloaddition intermediate, although additional transformations would be required to convert the resulting ketone into olefin moiety. Enolate 15 can be generated by a base-promoted deprotonation of cyclopentenone 16, which was presumably accessible through a series of functional group interconversions (FGIs) from a known guaiane-type 5-7-5 tricyclic precursor 17^32–34, readily available from the natural α-santonin (18) in just 3 steps of chemical transformations via the classic photo-induced santonin rearrangement. Notably, the use of the known compound 17 made the stereochemistry of C10 alcohol in 16 different from the intermediate 10 in the biogenetic proposal, which would not affect our synthetic study as a bioinspired dehydration would be implemented to erase this stereocenter and generate the olefin in artatrovirenol B (6).

Fig. 3. Retrosynthetic Analysis.

Bioinspired retrosynthetic analysis of artatrovirenols A and B.

Total syntheses of artatrovirenols A and B

Our synthesis commenced with the preparation of the known carbocyclic lactone 17 from α-santonin on large-scale preparation (Fig. 4). According to the literature with a slight modification, 17 could be readily obtained in 3 steps on decagram scale and sufficiently supply to the following explorations^32–34. Then, double carbonyl α-selenation and oxidative elimination in one pot generated two double bonds with high regioselectivity, yielding compound 19 in moderate yield. Subsequent lactone saponification provided the corresponding γ-hydroxyl acrylic acid intermediate with the loss of TMS protection of the C10 tertiary alcohol at the same time. The resulting γ-hydroxyl acrylic acid was then subjected to the chemoselective silylation of the C6 secondary alcohol. However, the acrylic acid was also silylated at the same time, which promoted us to conduct a base-promoted chemoselective desilylation with K₂CO₃ to afford 20 in one pot from 19 in 37% yield on large scale. The low overall yield came from the partial lactonization event in competition with the expected silylation in the TBS-protecting operation. To mask the active free carboxylic acid, esterification of 20 with excess diazo reagent TMSCHN₂ gave the corresponding methyl ester 16 in excellent yield, allowing us to explore the subsequent bioinspired formal [4 + 2] cycloaddition reaction.

Fig. 4. Investigation of the formal [4 + 2] cycloaddition reaction.

Reagents and conditions: (a) LiHMDS (2.2 equiv.), PhSeBr (3.0 equiv.), THF, −78 °C, 1 h, then, mCPBA (3.5 equiv.), CH₂Cl₂, −78 °C to RT, 12 h, 74%; (b) KOH (1.5 equiv.), THF/H₂O = 1/1, 20 min, then, TBSCl (4.0 equiv.), Imid. (8.0 equiv.), CH₂Cl₂, RT, 16 h, then, K₂CO₃ (1.0 equiv.), MeOH/THF = 1/1, RT, 1 h, 37%; (c) TMSCHN₂ (1.3 equiv.), benzene/MeOH 3.5/1, RT, 5 min, 93%; (d) TBD (2.0 equiv.), toluene, 120 °C, 30 min, 60%; (e) TsOH (2.0 equiv.), benzene, reflux, 1 h, 94%; (f) TBSOTf (1.2 equiv.), 2,6-lutidine (2.7 equiv.), CH₂Cl₂, −78 °C, 1 h, 92%. LiHMDS lithium bis(trimethylsilyl)amide, THF tetrahydrofuran, m-CPBA = 3-chloroperoxybenzoicacid, TBSCl tert-butyldimethylsilyl chloride, Imid. imidazole, MeOH methanol, RT room temperature, TMSCHN₂ = (trimethylsilyl)diazomethane, TBD = 1,5,7-triazabicyclo[4.4.0]dec-5-ene, p-TsOH p-toluenesulfonic acid, TBSOTf trifluoromethanesulfonic acid tert-butyldimethylsilyl ester, 2,6-lutidine 2,6-dimethylpyridine.

With the crucial bicyclic precursor 16 in hand, the bioinspired intramolecular [4 + 2] cycloaddition reaction to construct the key caged tetracyclic core was then investigated (Table 1). Although both acids^35–38 and bases^39–42 have been reported to promote such transformations in simple substrates, here in our case, the bicyclic structure of 16 containing diverse functional groups, particularly the C10 tertiary alcohol, might be incompatible with strong Lewis or Brønsted acids. Thus, we explored diverse types of bases. According to the previous observation in the conversion of 19 to 20 that base would lead to the structure decomposition, basicity would play a pivotal role in this reaction. First, when the cyclopentenone was deprotonated at low temperature by the common enolate-formation metal amide bases like LDA, NaHMDS and KHMDS, the in situ formed enolate 15 did not provide any cycloadduct product at all, even with slow increase to room temperature (RT) (entries 1‒3). Inorganic base tert-BuOK did not change this result at RT (entry 4). Considering the tertiary C4 position is bulky to be deprotonated and the subsequent [4 + 2] cycloaddtion requires a highly congested conformation, we then tried at higher reaction temperature. Encouragingly, the expected [4 + 2] cycloaddition product 21 was successfully obtained with full conversion, when the reaction was run at 70 °C with tert-BuOK as base in THF, albeit in low yield (28%) (entry 5). The structure of the obtained 21 was unambiguously determined by x-ray diffraction analysis of the acid-promoted dehydrated and desilylated derivative 22 (CCDC 2272312). Then, when toluene was used as the solvent, the reaction did not occur at 70 °C and proceeded quite slowly at 80 °C. Higher temperature (120 °C) increased the yield a little bit (entry 6). Subsequent investigations on other bases increased the yield up to 60% by screening organic bases DBU and TBD (entries 7, 8). However, further increasing the temperature in xylene solution did not give better results, probably due to the material decomposition (entries 9, 10). Thus, TBD provided proper basicity in this reaction to balance reactivity and decomposition aspects. To further re-check the role of temperature with TBD as the base, the reaction was run at RT again, and no reaction was observed, suggesting the cycloaddition is indeed a thermo-induced process (entry 11). Up till now, the critical highly compacted three-dimensional tetracyclic cage of the target molecules has been rapidly forged from starting materials with the challenging three bridge-head all-carbon quaternary centers been properly created in a single step. Considering this bioinspired [4 + 2] cycloaddition reaction itself is of great challenge to proceed and it, significantly, could be performed on large scale (entry 8), the optimal TBD/toluene system was believed to be capable of supplying enough material to the following synthetic explorations. Mechanistically, for the possible reaction pathway of this formal [4 + 2] cycloaddition between the enone-derived enolate and acrylate moieties, such base-promoted transformations were generally regarded as a stepwise double Michael reaction^19,39–42.

Table 1.

Optimization of the formal [4 + 2] cycloaddition reaction.^a

Entry	Base	Solvent	T (°C)	Yield (%)b
1	LDA	Et2O	−78 to RT	NR
2	NaHMDS	THF	−78 to RT	NR
3	KHMDS	THF	−78 to RT	NR
4	tert-BuOK	THF	RT	NR
5	tert-BuOK	THF	70	28
6	tert-BuOK	Toluene	120	35
7	DBU	Toluene	120	31
8	TBD	Toluene	120	60 (60c)
9	DBU	Xylene	150	41
10	TBD	Xylene	150	56
11	TBD	Toluene	RT	NR

^aAll reactions were performed with 16 (0.2 mmol) and base (0.4 mmol) in solvent (2 mL) at the corresponding temperature.

^bIsolated yields are given.

^cRun on 650 mg scale. NR no reaction, LDA lithium diisopropylamide, NaHMDS sodium bis(trimethylsilyl)amide, KHMDS potassium bis(trimethylsilyl)amide, tert-BuOK potassium tert-butoxide, DBU = 1,5-diazabicyclo[5.4.0]undec-5-ene.

Having constructed the critical caged tetracyclic scaffold, we moved on to finish the synthesis of artatrovirenol B (6) through a series of FGIs. Since intermediate 22 already contains the required endocyclic C9–C10 double bond, it was directly used for the next synthetic studies. However, intensive efforts to convert the C3 ketone into endocyclic C2‒C3 double bond all failed through the approaches including, but not limited to, carbonyl reduction/dehydration^43–46, tosyl hydrazone formation/base-promoted Shapiro olefination⁴⁷, and enol triflate formation/Pd-catalyzed reduction (see Figure S5 in the Supplementary Information), no matter with 22 or its TBS ether 23^48,49. Presumably, these failed results suggest that the existence of C9‒C10 olefin makes the caged system highly rigid and strained, thus prone to proceed a ring-opening process to release the strain under the above-mentioned conditions. Thereby, we postponed the employment of C9‒C10 olefin and paid the attention to firstly construct the C2‒C3 olefin from intermediate 21 with saturated C9‒C10 bond.

As shown in Fig. 5, protection of the C10 tertiary alcohol in 21 with TBS followed by C3 ketone reduction with Corey-Bakshi-Shibata (CBS) reaction afforded alcohol 24 with exclusive stereoselectivity. The C3 alcohol configuration was determined by the nuclear Overhauser effect (nOe) experiment. Manipulation of alcohol 24 with common dehydration methods, such as Burgess reagent, Martin sulfurane, Brønsted acids, and mesylation/elimination, all failed again to give the desired C2‒C3 olefin. Inspired by Zhu’s work²⁹, Chugaev elimination was then explored. Alcohol 24 was efficiently transformed into the corresponding thiocarbonate 25, which subsequently underwent a thermo-induced elimination process at remarkably high temperature (220 °C) to generate the expected C2‒C3 olefin in product 26. Notably, the stereochemistry of C3-hydroxy group is important to the elimination reaction. Initially, the carbonyl group at C3 position was reduced with sodium borohydride to give two separable diastereomers in a 3:1 ratio, in which the minor one was elaborated to C3-epi-25 and this epimer decomposed in the elimination reaction. So, the enantioselective reduction method by (R)-CBS was used to provide a solely stereoselective reduction of the carbonyl group at C3 position. Gratifyingly, the following C9‒C10 olefin could be successfully formed just by adding TsOH into the reaction system to promote the desilylation/dehydration transformations, and moreover, the final ester saponification was capable of giving artatrovirenol B (6) with excess KOH at elevated temperature. These three transformations could be performed in one pot from 25 in acceptable overall yield. The obtained synthetic sample of 6 showed identical physical data with the natural one. Thus, we have achieved the synthesis of artatrovirenol B (6) in an asymmetric fashion through a bioinspired approach in 8 steps from the readily available known precursor 17.

Fig. 5. Completion of the synthesis of artatrovirenols A and B.

Reagents and conditions: (a) TBSOTf (2.4 equiv.), 2,6-lutidine (5.4 equiv.), CH₂Cl₂, 0 °C to RT, 2 h, (b) (R)-CBS (0.3 equiv.), BH₃·THF (2.0 equiv.), THF, 0 °C, 15 min, 92% (over 2 steps), (c) KHMDS (1.5 equiv.), PhOC(S)Cl (3.0 equiv.), THF, –78 °C, 15 min, 98%, (d) Ph₂O, 220 °C, 10 h, then, TsOH (2.0 equiv.), benzene, 100 °C, 30 min, then, KOH (15.0 equiv.), MeOH/H₂O = 31, 80 °C, 48 h, 40%, (e) VO(acac)₂, (0.5 equiv.), TBHP (3.0 equiv.), DCE, 40 °C, 30 min, 48%. (R)-CBS = (R)−3,3-diphenyl-1-methylpyrrolidino[1,2-c]−1,3,2-oxazaborole, VO(acac)₂ vanadyl acetylacetonate, TBHP tert-butyl hydroperoxide, DCE = 1,2-dichloroethane.

Next, the biomimetic conversion of artatrovirenol B into A through epoxidation and lactonization process was investigated. Pleasingly, treating 6 with a catalytic amount of VO(acac)₂ and excess oxidant TBHP at RT successfully gave artatrovirenol A (5), accompanied with its C9-epimer byproduct in a 1:1 ratio. Notably, a slight increase of temperature to 40 °C dramatically accelerated the reaction rate. This result suggests that the olefin epoxidation displayed no facial selectivity and the subsequent epoxide-opening lactonization could occur smoothly under this acidic condition. To tackle the diastereoselectivity problem, we supposed that the C6 alcohol could provide directing effect to drive the oxygen atom to approach from the upper face of the olefin through the reaction model 27. This hypothesis was achieved in practice by reacting VO(acac)₂ catalyst with substrate 6 at 40 °C for 30 min to make the vanadyl center bonded with the C6 alcohol through ligand exchange prior to adding the oxidizing reagent TBHP to facilitate the subsequent epoxidation and lactonization events. Finally, artatrovirenol A (5) was generated as a sole stereoisomer, albeit in 48% yield. The absolute configurations of synthetic 5 were further confirmed by x-ray crystallographic analysis (CCDC 2363780), and the optical rotation data comparison determined the absolute configurations of natural artatrovirenol A (5). The phenomenon that the lactonization proceeded smoothly as soon as the epoxidation occurred serves as a solid evidence to support that artatrovirenol B (6) might be a biogenetic precursor of artatrovirenol A (5) in nature. Thus, a 9-step route to access 5 from 17 is developed, which is a comparably concise approach for synthesis of artatrovirenol A (5), compared to Zhu’s de novo synthesis.

Discussion

In conclusion, we have achieved the total synthesis of artatroverinol B in 8 steps starting from a known bicyclic guaiane-type precursor 17, based on a bioinspired formal [4 + 2] cycloaddition strategy, and further realized its biomimetic transformation to artatroverinol A through an epoxidation-mediated lactonization protocol. This research provides a concise approach to the asymmetric synthesis of both artatroverinols A and B in a fully stereocontrolled fashion, and on the other hand, serves as a solid evidence to support the proposed biogenetic pathway in which the complex caged tetracycle might be formed through an IMDA process from the planar guaiane-type sesquiterpenoid and artatrovirenol B might be a biogenetic precursor of artatrovirenol A. The developed bioinspired [4 + 2] cyclization strategy is supposed to be applicable for the synthesis of daphnenoid A, which is ongoing in our laboratory.

Methods

All non-aqueous reactions were carried out using oven-dried glassware under a positive pressure of dry argon unless otherwise noted. All reagents and solvents were reagent grade. Further purifications and drying by standard methods were used when necessary. Except as indicated otherwise, reactions were magnetically stirred and monitored by thin layer chromatography (TLC) using Merck Silica Gel (60F-254) plates and visualized by fluorescence quenching under UV light. In addition, compounds on TLC plate were visualized with a spray of 5% w/v phosphomolybdic acid (PMA) in ethanol and with subsequent heating. Chromatographic purification of products (flash chromatography) was performed on E. Merck Silica Gel 60 (230−400 mesh). All evaporation of organic solvents was carried out with a rotary evaporator. Yields refer to chromatographically and spectroscopically pure compounds, unless otherwise stated. The photochemical reaction was irradiated with a high-pressure mercury lamp (400 W, equipped with a quartz water cooling jacket), and cooled by an electric fan to maintain at 23–25 °C. ¹H NMR and ¹³C NMR spectra were recorded in CDCl₃ solution on 400 M or 600 M instrument as stated. Chemical shifts for ¹H NMR and ¹³C NMR spectra were reported in ppm (δ) (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad signal) and were referenced to either TMS (δ_H = 0.00 ppm, δ_C = 0.0 ppm) or residual undeuterated solvent as internal standard (CDCl₃: δ_H = 7.26 ppm, δ_C = 77.0 ppm). High-resolution mass spectra (HRMS) were recorded on a Thermo Scientific Orbitrap Exploris 120 mass spectrometer using an electrospray ionization (ESI) technique and a TOF analyzer. X-ray diffraction analysis of single crystals was performed on Bruker APEX II X-ray single crystal diffraction meter. Melting points were measured on a melting point apparatus, which was not calibrated before use. Optical rotations were obtained on 0.1 mL cell with a 1 cm path length on Rudolph Autopol IV automatic polarimeter and concentrations (c) were reported in g × (100 mL)⁻¹.

Author contributions

Y.Y., X.X., and X.S. conceived the project. Y.Y., L.H., Y.L., and J.D. conducted all the experimental work and analysed the data. D.X. and P.-P.Z. analysed the data. The paper was written by H.L., X.X., and X.S. with proofreading from all authors. X.X. and X.S. directed the project.

Peer review

Peer review information

Nature Communications thanks Jijun Chen and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support the findings of this study are available within the article and its Supplementary Information. The X-ray crystallographic data for structures reported in this study have been deposited at the Cambridge Crystallographic Data Centre (CCDC) under deposition numbers 2272312 (22), 2363780 (5). These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_request@ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 33603. All data are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-55560-9.