Enantioselective Intermolecular [2+2] Photocycloaddition Reaction of Cyclic Enones and Its Application in a Synthesis of (−)-Grandisol

Abstract

The intermolecular [2+2] photocycloaddition of typical cyclic α,β-unsaturated enones, such as 2-cyclohexenone, with olefins was performed in moderate to good yields (42–82%) and with high enantioselectivity (82%–96% ee). An unusual substitution pattern at the chiral oxazaborolidine-AlBr₃ Lewis acid complex that promotes the reaction was found to be crucial for the success of the reaction. The method was applied to the enantioselective synthesis of the monoterpene (−)-grandisol, which could be accomplished in six steps and with an overall yield of 13% starting from 3-methyl-2-cyclohexenone.

Although the intramolecular [2+2] photocycloaddition reaction of cyclic enones had been discovered as early as 1908,¹ it was not before the 1960s that the first intermolecular variants of this reaction were reported.² Over the last 40 years, the reaction has been employed as a key step in literally hundreds of natural product syntheses³ and it represents one of the most powerful transformations of organic chemistry.⁴ Typically, the reaction is induced by direct excitation of an enone, such as 2-cyclohexenone (1a), the triplet state (T₁) of which is populated by rapid intersystem crossing (ISC) from the first excited singlet state (S₁). The former state is long-lived and can be captured by a photochemically inactive alkene to generate a cyclobutane. A representative reaction is the addition of isobutene, which was employed in Corey’s landmark synthesis of racemic caryophyllene (Scheme 1).⁵

Scheme 1. Corey’s Landmark Synthesis of Racemic Caryophyllene with an Intermolecular [2+2] Photocycloaddition Reaction of 2-Cyclohexenone (1a) as the Key Step.

Although the olefin is commonly used in excess, the reaction is not perfectly chemo- and regioselective and an extensive purification is required to obtain product rac-2a in yields of 35–45%. The treatment with basic alumina facilitates the epimerization of the stereogenic center in α-position to the ketone carbonyl group but does not remove any undesired constitutional isomers. Most severely, there is up to this date no way to access the cyclobutane products of a typical intermolecular enone [2+2] photocycloaddition reaction in an enantioselective fashion.^6,7 Indirect methods that rely on diastereoselective reactions require a major synthetic effort involving multistep sequences.⁸ Recently developed protocols for enantioselective intermolecular [2+2] photocycloaddition reactions⁹ are not applicable to unfunctionalized cyclic enones, such as 2-cyclohexenone (1a). We have now found a way to prepare cyclobutanes such as 2a from enones with high enantioselectivity and our results are disclosed in this communication.

In preliminary work, we studied the UV/vis properties of 2-cyclohexenone (1a) upon treatment with Lewis acids (Figure 1). In line with previous observations,^10,11 the strong ππ* absorption of the parent compound at λ = 224 nm (ε = 13 400 M^–1 cm^–1) was shifted bathochromically and was recorded at λ = 260 nm (ε = 9670 M^–1 cm^–1) with EtAlCl₂ and at λ = 266 nm (ε = 13 080 M^–1 cm^–1) with the stronger Lewis acid BCl₃. In both cases, there is still a detectable absorption of the complex at λ = 330 nm where the weak (ε = 70 M^–1 cm^–1) nπ* absorption of uncomplexed 2-cyclohexenone is located. Because it was expected that the AlBr₃-based oxazaborolidine Lewis acids¹² would exert an even stronger bathochromic shift than BCl₃, it seemed possible to selectively excite the complex of cyclohexenone with a chiral Lewis acid but not the uncomplexed substrate. In this scenario, enantioselective intermolecular [2+2] photocycloaddition to an alkene should be possible as previously found for intramolecular reactions.^10,13 However, several questions remained open, e.g., whether the lifetime of the complex would be sufficiently long to allow for enantioface differentiation before dissociation and whether previously used Lewis acids would provide a reasonable enantioselectivity. It was also unclear whether the photoexcited cyclohexenone would lead to a destruction of the Lewis acid and whether the known side reactions of intermolecular [2+2] photocycloaddition reactions would be enhanced. Initial results with the previously used^13b Lewis acid 3a seemed to confirm the worst concerns (Table 1, entry 1). The enantioselectivity in the reaction of 1a with 2-ethylbut-1-ene under typical conditions was disappointingly low and the Lewis acid seemed unstable under the reaction conditions. Variation of Ar’ showed that the insufficient performance was not associated with this substituent but rather with the aryl group Ar (entry 2). Gratifyingly, an extensive screening of oxazaborolidines¹⁴ with different substituents Ar delivered Lewis acids with which the selectivity issues could be smoothly overcome. The superior substituent was a 2,3-dimethylphenyl group that was incorporated for example in Lewis acids 3c–3e.

Figure 1.

UV/vis spectra of 2-cyclohexenone (1a) in the absence (—) and in the presence of 20 equiv of either EtAlCl₂ (− – −) or BCl₃ (- - -) (c = 0.5 mM in CH₂Cl₂).

Table 1. Enantioselective Intermolecular [2+2] Photocycloaddition of 2-Cyclohexenone (1a) and 2-Methyl-1-butene Mediated by Chiral Lewis Acids 3.

^aThe crude product was purified by chromatography on silica gel (eluent: pentane/Et₂O = 4/1). The material was dissolved in CH₂Cl₂ and successively treated with basic alumina and dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate. Filtration through a short pad of silica and solvent removal delivered pure compound 2b.

^bThe enantiomeric excess was determined by chiral GLC analysis.

The 2,3,4-trifluorophenyl derivative 3c gave already a much improved yield and enantioselectivity (entry 3) compared to 3b. The performance was further improved when employing a 2,4,6-trifluorophenyl group as Ar’ (entry 4). The last entry of the table (entry 5) illustrates that also the aryl group Ar’ has a subtle influence on the enantioselectivity but not on the regio- and chemoselectivity. Indeed, apart from the high enantioselectivity, it is notable that the yield of the [2+2] photocycloaddition significantly increased compared to the racemic reaction (see Supporting Information), which delivered a yield of only 51% for product rac-2b.

Lewis acid 3d evolved from the optimization studies as the optimal choice to promote an enantioselective intermolecular [2+2] photocycloaddition of cyclic enones 1. The Lewis acid was prepared in situ by addition of AlBr₃ to the respective oxazaborolidine and it was applied to a variety of enone/olefin combinations (Table 2). The reason for the high catalyst loading of 50 mol % is the racemic background reaction that can only be suppressed if the catalyst concentration is high.^10b

Table 2. Enantioselective Intermolecular [2+2] Photocycloaddition of Cyclic Enones (1) and Various Terminal Olefins to Bicyclic Products 2^a.

^aReactions were performed at a concentration of c = 20 mM in CH₂Cl₂ solution. Workup included treatment with basic alumina in CH₂Cl₂ (see Table 1).

^bThe purified material was treated with dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate to remove olefinic impurities.

^cThe reaction remained incomplete and the yield is based on enone conversion.

The Lewis acid catalyzed process enabled an enantioselective access to cyclobutane 2a, which was obtained in higher yield than racemic compound rac-2a (Scheme 1). It is known that the levorotatory enantiomer ([α]_D²³ = −152.8 at c = 1.4 in CHCl₃) ent-2a exhibits the (1R,6R)-configuration at the two stereogenic centers of the bicyclo[4.2.0]octane skeleton.¹⁵ Product 2a was dextrorotatory ([α]_D²⁵ = +162.6 at c = 1.4 in CH₂Cl₂) proving unambiguously its absolute configuration as (1S,6S). The absolute configuration of the other products 2 was assigned in analogy and matches our model for the enantioface differentiation (vide infra). Product yields were moderate to good (42–82%) and the enantioselectivity was consistently high (82%–96% ee). Olefinic substrates were restricted to ethylene (products 2h, 2i, 2l, 2m, 2o, 2p) and to symmetric 1,1-disubstituted olefins in order to avoid the formation of cis/trans-diastereoisomers.¹⁶ Oxygen and chlorine substituents were tolerated (products 2e, 2f, 2g, 2n) and several cyclobutanes were prepared that had previously been employed in racemic form as starting material for natural product total syntheses.^5,17 Enone substitution was varied at the β-position (products 2l, 2m, 2n, 2p) and within the six-membered ring of 2-cyclohexenone (products 2i, 2j, 2k, 2m, 2n). Five-membered substrates reacted with high enantioselectivity (products 2h, 2o) but the reaction was slower than the reaction of the respective 2-cyclohexenones.

In order to showcase the synthetic utility of the enantioselective intermolecular [2+2] photocycloaddition reaction we performed a total synthesis of (−)-grandisol (Scheme 2). Grandisol is a component of the aggregation pheromone of the cotton boll weevil (Anthonomus grandis) and has received a lot of attention by the synthetic community.¹⁸ Although many syntheses exist and some of them provided enantiomerically enriched product,¹⁹ there is no synthesis that features an enantioselective [2+2] photocycloaddition as the key step. Our synthesis follows a known bond set^19f and commenced with photocycloaddition product 2l. The introduction of the olefinic double bond was accomplished by the catalytic Saegusa oxidation protocol reported by Stahl and co-workers.²⁰ Addition of methyl lithium to enone 4 produced tertiary alcohol 5 the enantiopurity of which could be nicely increased by recrystallization. Oxidative cleavage of the double bond and further oxidation to keto acid 6 was performed with sodium periodate and RuCl₃ as the catalyst. Olefination to product 7 and the reduction of the acid to the alcohol proceeded uneventfully and delivered (−)-grandisol in 96% ee. Starting from 3-methyl-2-cyclohexenone, the synthesis proceeded in only six steps with an overall yield of 13% and thus represents a concise route to the enantiopure natural product.

Scheme 2. Enantioselective Total Synthesis of (−)-Grandisol (8).

Mechanistic considerations concerning the enantioselective intermolecular [2+2] photocycloaddition, which go beyond previously established facts,¹⁰ must account for the improved performance of Lewis acid 3d compared to 3a. So far, only intramolecular Lewis acid-catalyzed [2+2] photocycloaddition reactions had been studied with substrates (dihydropyridones, β-alkoxyenones) that bear a tethered olefin in β- or γ-position. Assuming a reaction on the triplet hypersurface,²¹ the first bond formation step in these cases occurs rapidly in the β-position. Intermolecular trapping of photoexcited enones, such as 2-cyclohexenone (1a), in a Lewis acid complex is likely slower and it occurs with high preference in α-position (vide infra). The former fact, possibly in combination with the reactive T₁ state of 1a, could be responsible for hydrogen abstraction²² in complex 1a·3a thus leading to a decomposition of the catalyst. In complex 1a·3d, the methyl groups are oriented differently (e.g., as shown in Figure 2) and not only escape hydrogen abstraction but also lead to an improved enantioface differentiation in α-position. Bond formation occurs from the Si face and establishes the first stereogenic center at the future carbon atom C₁ of the bicyclo[4.2.0]octane skeleton. Circumstantial evidence for the intermediacy of 1,4-diradicals such as 9 stems from the observation that olefinic products are formed as minor impurities in the reactions with 1,1-disubstituted alkenes. They arise likely from hydrogen abstraction, which is a side reaction competing with C–C bond formation. Synthetically, the occurrence of the byproducts is a minor issue as they can be removed by treating the crude reaction mixture with dimethyl 1,2,4,5-tetrazine-3,6-dicarboxylate²³ (see Tables 1 and 2).

Figure 2.

Suggested structure of 1:1 enone/Lewis acid complexes 1a·3a and 1a·3d and of putative 1,4-diradical intermediate 9.

The improved regioselectivity of the Lewis acid-catalyzed intermolecular [2+2] photocycloaddition is associated with the above-mentioned preference of α-attack, which in turn is evidenced by the exclusive formation of head-to-tail photocycloaddition products. No head-to-head products were identified in the crude product mixture while they are formed in notable amounts in the course of racemic [2+2] photocycloaddition reactions. It appears as if the Lewis acid increases the polarity of the excited state which is opposite to the ground state polarity.⁴ In the ground state the β-position of an enone is electrophilic, whereas in the excited state the α-position is electrophilic. In a simplistic picture this polarity reversal is due to the ππ* character of the triplet state with the half-filled π orbital lacking electron density in α-position.

In summary, the substitution pattern of the oxazaborolidine was found to have a strong influence on the outcome of the enantioselective intermolecular [2+2] photocycloaddition of enones which was mediated by an oxazaborolidine-AlBr₃ complex. Catalyst 3d evolved as the catalyst of choice enabling high enantioface differentiation (82%–96% ee) and good chemoselectivity. Studies are ongoing to reveal its mode of action and to detect potential reaction intermediates.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.8b01011.

• Experimental procedures, analytical data and NMR spectra for all new compounds, GLC traces of products (PDF)

• NMR spectra for all new compounds (ZIP)

The authors declare no competing financial interest.