Concise Access to the Skeleton of Protoilludane Sesquiterpenes through a Photochemical Reaction Cascade: Total Synthesis of Atlanticone C

Abstract

In a single photochemical operation (λ≥350 nm) an easily accessible indanone derivative was converted into a structurally complex precursor of the protoilludane sesquiterpenes. The product (60 % yield) contains all 15 carbon atoms of the skeleton in the required connectivity and was transformed into the natural product atlanticone C (9 steps, 6 % overall yield). In addition, it was shown that other protoilludanes, such as Δ⁶‐protoilludene and paesslerin A, can be prepared in a concise fashion via the photochemical key intermediate. The photochemical reaction cascade comprises an ortho photocycloaddition, a thermal disrotatory ring opening and a regioselective disrotatory [4π] photocyclization.

Magic moments: Upon irradiation, the easily accessible indanone 1 (4 steps, 57 %) was converted in a single step into complex tetracyclic product 2, which contains all of the carbon atoms of the protoilludane skeleton. Compound 2 served to synthesize atlanticone C (3), and was shown to be a useful intermediate for the synthesis of other protilludanes.

The protiolludane skeleton A (Figure 1) represents an intriguing natural product scaffold and is the hallmark of many biologically active sesquiterpenes. A central cyclohexane ring is flanked by a cyclobutane ring with a quaternary carbon center C‐3 and by a gem‐dimethyl‐substituted cyclopentane ring at positions C‐2 and C‐9. The protoilludane biosynthesis ^[1] from farnesyl pyrophosphate via cationic intermediates is testimony to the masterful ability of nature to create molecular complexity from simple building blocks. Representative sesquiterpenes with this structure element include atlanticone C (1), ^[2] Δ⁶‐protoilludene (2), ^[3] and paesslerin A (3). ^[4]

Figure 1.

The protoilludane skeleton A and representative protoilludane sesquiterpenes: atlanticone C (1), Δ⁶‐protoilludene (2), paesslerin A (3).

Many synthetic approaches have been described to tackle the challenging core structure of the protoilludanes. ^[5] One set of syntheses aimed at a construction of the tricyclic skeleton from acyclic or monocyclic precursors in a manner mimicking their biological genesis from humulene. ^[6] In several other approaches, the rings were built consecutively, ^[7] with the construction of the cyclobutane ring requiring particular attention. Notable photochemical strategies ^[8] to secure formation of the four‐membered ring include the common [2+2] photocycloaddition ^[9] but also some less frequently used methods such as the photochemical 1,3‐acyl shift reaction of β,γ‐unsaturated enones. ^[10] Our interest in protoilludanes was triggered by recent work ^[11] on a reaction cascade ^[12] that is initiated by an ortho photocycloaddition and that generates the bicyclo[4.2.0]octane skeleton, a core element of structure A, in a single operation. Herein, we disclose a concise access to protoilludane sesquiterpenes that culminated in the first total synthesis of atlanticone C (1) and in the stereoselective synthesis of two other advanced intermediates that had been previously converted into Δ⁶‐protoilludene (2) and paesslerin A (3).

Arenes B (Scheme 1), which have a carbonyl group in ortho position to an alkenyloxy group, are known to undergo an initial [2+2] photocycloadditon at the arene (ortho photocycloaddition[ ¹³ , ¹⁴ ]). Products C are unstable and form cyclooctatrienes D through a disrotatory ring opening. In some cases, these products have been isolated ^[15] but the most frequent reaction pathway is a consecutive disrotatory [4π] cyclization, which can lead to products E or E′ depending upon the substitution pattern within the tether. ^[16]

Scheme 1.

A Photochemical reaction cascade triggered by an ortho photocycloaddition of salicylic acid derivatives B to give the constitutional isomers E and E′ (gray circles indicate the carbon atoms of the former olefinic double bond).

Following up on earlier work by Wagner and co‐workers, ^[17] we have recently found that several salicylates and salicyclic acid derivatives deliver clean formation of products E if irradiated at λ=300 nm in methanol solution. ^[11] Given the structural similarity of E with the biyclo[4.2.0]cyclooctane part of structure A, we reasoned that a substituent Y=O would facilitate a ring opening of the former tether and would secure the introduction of the hydroxy group required at position C‐13 of the natural product atlanticone C. Regarding the other variable R in structure E, a late‐stage introduction of the C‐12 methyl group did not seem feasible. It was thus mandatory to introduce this substituent already in the starting material, that is, into the meta position of the salicylate. The respective salicylic acid is readily available through a Kolbe–Schmitt reaction, ^[18] and substrate 4 b (Scheme 2) was synthesized in three steps and an overall yield of 77 % from para‐cresol (see the Supporting Information). The results of its photochemical reaction were disappointing, however. While the unsubstituted salicylate 4 a had given the desired product 5 a in 65 % yield in previous work, ^[11] substrate 4 b delivered an inseparable mixture of diastereoisomers in only 46 % yield under identical reaction conditions with a strict preference for the opposite regioisomer. Attempts to alter this reaction outcome by changing the light source or the solvent remained unsuccessful. Apparently, a larger R group (R=Me instead of R=H) avoids the sterically encumbered position at the quaternary carbon center and the [4π] ring closure to the linear product 5 b′ becomes competitive.

Scheme 2.

Photochemical reaction of methyl salicylates 4 to give regioisomeric products 5 and 5′ (r.r.=regioisomeric ratio).

Inspection of molecular models indicated that a cyclic carbonyl compound, that is, a ketone, might enforce the desired reaction pathway, delivering a product E with an even higher structural similarity to skeleton A. However, there was no precedence for the ortho photocycloaddition of cyclic aromatic ketones, and many photochemical reaction pathways are conceivable for the keto group both in a potential substrate and its product. Despite these concerns, we decided to investigate this topic more closely, and ketone 6 (Scheme 3) was easily synthesized from para‐methyl anisole in four steps and an overall yield of 57 % (see the Supporting Information). Gratifyingly, this substrate could be successfully taken into the desired photochemical reaction cascade without notable decomposition. Careful optimization eventually led to conditions under which the desired product 7 was reproducibly formed in yields of 60 %. In order to avoid undesirable consecutive reactions of enone 7 an Fe₂(SO₄)₃ solution (c=300 mg L⁻¹) in 0.01 m aqueous HCl ^[19] was employed to filter the desired wavelength region from fluorescent lamps emitting at a maximum of λ=350 nm. ^[20] Parallel to extensive NMR analyses the structure of crystalline diastereomerically pure product 7 was secured by X‐ray crystallography. ^[21]

Scheme 3.

Successful photochemical reaction cascade of ketone 6 to tetracyclic product 7, the structure of which was elucidated by single crystal X‐ray crystallography.

The transformation of ketone 7 to atlanticone C (1) required the adjustment of the oxidation state at several carbon atoms and the introduction of the stereogenic center at carbon atom C‐2. The tetracyclic skeleton directs an attack of a given reagent to the bottom face, which made it impossible to secure the relative configuration at C‐2 by a direct reduction. After hydrogenation of the cyclobutene to a cyclobutane ring, it was therefore attempted to employ the functional group at position C‐10 in intermediate 8 (Scheme 4) as a directing group. Ketone reduction was diastereoselective but subsequent attempts to perform a directed hydrogenation of the allylic alcohol ^[22] remained unsuccessful. Instead, the ketone was converted into hydrazone 9 by condensation ^[23] with N‐tosylhydrazine (58 % yield, 36 % of unreacted starting material). Subsequent reduction according to the method of Kabalka et al. ^[24] not only delivered the hydride from the correct diastereotopic face via intermediate 10 but also established the double bond in product 11 at the desired position C‐9/C‐10.

Scheme 4.

Synthesis of atlanticone C (1), exact conditions and yields: a) H₂ (1 atm), 10 mol % Pd/C, (EtOAc), rt, 1 h, 99 %; b) H₂NHNTs (1.5 equiv.), 10 mol % H₂SO₄, (MeOH), 65 °C, 24 h, 58 % (91 % based on conversion); c) catecholborane (4.0 equiv.), HOAc (8.0 equiv), (CHCl₃/THF), −40 °C, 3 h, then NaOAc⋅3 H₂O (16 equiv.), −40 °C→52 °C, 16 h, 92 %; d) BuLi (10 equiv.), KO^tBu (10 equiv.), (THF/hexane), −78 °C, 1 h, then BF(OMe)₂⋅OEt₂ (16.6 equiv.), 5 min, then 10 m NaOH (100 equiv.), 30 % H₂O₂ (100 equiv.), −78 °C→rt, 18 h, 87 %; e) (COCl)₂ (2.0 equiv.), DMSO (5.0 equiv.), (CH₂Cl₂/DMSO) −78 °C, 15 min; NEt₃ (10 equiv.), −78 °C→−40 °C, 30 min, short work‐up (see Supporting Information), then MsCl (7.25 equiv.), DMAP (3.63 equiv.), H₂O (2.9 equiv.), (CH₂Cl₂), rt, 3 h, 68 % over two steps; f) Dibal‐H (1.2 equiv.), (CH₂Cl₂), −78 °C, 1 h, 93 %; g) Ac₂O (5.0 equiv.), 10 mol % DMAP, (py), rt, 30 min, 65 %; h) py (75 equiv.), CrO₃ (20 equiv.), (CH₂Cl₂), rt, 3 h, 44 %; i) K₂CO₃ (8.0 equiv.), (MeOH/H₂O), rt, 40 min, 68 %. Ts=para‐toluenesulfonyl, Ms=methanesulfonyl, DMAP=4‐dimethylaminopyridine, Dibal‐H=diisobutylaluminum hydride.

In order to adjust the oxidation state at positions C‐6, C‐7, and C‐8, the 1,3‐dioxane ring of intermediate 11 had to be opened. Reported procedures for the ring opening of methylene acetals ^[25] require acidic conditions, which turned out to be not perfectly compatible with the substrate. For example, the yield of deprotection with trifluoroacetic acid anhydride and acetic acid in CH₂Cl₂ ^[26] was only 36 %. We therefore turned to a somewhat unusual method that makes use of the relative high acidity of the methylene protons and of the high stability of substrate 11 towards base. Treatment with the Schlosser–Lochmann base BuLi/KO^tBu and a subsequent quench with BF(OMe)₂⋅OEt₂ ^[27] delivered product 13 after oxidative work‐up. It is likely that the deprotection proceeds via boronate 12, which upon oxidative hydrolysis splits into diol 13 and formic acid. Since several known dehydration methods were not applicable to diol 13, we followed a procedure reported by Furukawa et al., ^[9d] according to which a 6‐hydroxy C‐13 protoilludane aldehyde can be dehydrated by treatment with MsCl, water, and DMAP. Reduction of aldehyde 14 to alcohol 15 and acetylation to ester 16 proceeded uneventfully. However, the allylic acetate 16 proved to be a labile compound, presumably due to its high electophilicity. In this regard, it was fortunate that oxygenation at the C‐8 methylene group could be executed with pyridine and CrO₃. Saponification of acetate 17 resulted in the formation of the desired alcohol 1. The spectroscopic data of synthetic product 1 perfectly matched the data reported for atlanticone C ^[2] and thus support the structural assignment of this natural product (see the Supporting Information for a comparison).

Compounds 18 and 19 (Scheme 5) were prepared as an additional exercise to illustrate the broad applicability of the photochemical reaction cascade to protoilludane synthesis. Compound 18 had been reported previously in the context of the total synthesis of paesslerin A (3) and had been obtained in 15 steps and 2 % yield from 5,5‐dimethylcyclopent‐2‐enone. ^[4b] Compound 19 had been described as a precursor of Δ⁶‐protoilludene (2) and had been prepared from methyl 4,4‐dimethyl‐2‐oxocyclopentane‐1‐carboxylate in 19 steps and 4 % yield. ^[9d]

Scheme 5.

Diastereoselective conversion of protoilludane precursor 13 into products 18 and 19, which have previously served as intermediates in the total synthesis of natural products 2 and 3.

The starting material for our synthetic approach was diol 13, which could be diastereoselectively hydrogenated to establish the stereogenic center at carbon atom C‐9 and deliver product 18. The facial diastereoselectivity is in line with previous work on Pt‐catalyzed hydrogenation reactions. ^[28] It is likely that the concave structure of the hexahydro‐1H‐indene directs the attack cis to the existing stereogenic center at C‐2. The overall yield for product 18 starting from para‐methylanisole, the precursor of ketone 6 (see above), was 13 % over 9 steps. Dess–Martin oxidation of alcohol 18 delivered ketone 19 in an overall yield of 8 % over 10 steps.

In conclusion, we have discovered a concise access to protoilludane sesquiterpenes from a substituted indanone. The reaction cascade establishes the correct constitution of the carbocyclic skeleton and also secures its relative configuration. In order to access more complex protoilludanes, it will be necessary to install further functional groups in the starting material or in the products. It is envisaged, for example, that the cyclobutene ring of photoproduct 7 could be employed to install hydroxy groups at positions C‐4 and C‐5, which are frequently encountered in protoilludanes. ^[29] Another intriguing topic relates to the control of the absolute configuration, which could be achieved either by starting from chiral enantiopure substrates or through enantioselective photochemical reactions. ^[30]

Conflict of interest

The authors declare no conflict of interest.

Supporting information

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.

Supplementary

A. Zech, C. Jandl, T. Bach, Angew. Chem. Int. Ed. 2019, 58, 14629.