Influence of Ring Strain on the Formation of Rearrangement vs Cyclization Isotwistane Products in the Acyl Radical Reaction of Bicyclo[2.2.2]octanone

An acyl radical reaction of bicyclo[2.2.2]octenone to yield either rearranged or cyclized isotwistane products is described. The influence of ring strain on the reaction was demonstrated by alternating the sizes of the fused ring in the starting material. DFT calculations showed that the reaction is under thermodynamic control and proceeds via a 5-exo-trig cyclization intermediate, which undergoes either hydrogen-atom transfer (HAT) to give a cyclized product or rearrangement via a twistane intermediate to give a rearranged product.

T he isotwistane motif, depicted in Scheme 1, is a rigid skeleton commonly found in various natural products, including the palhinines. 1,2Various strategies have been applied to its synthesis, such as Diels−Alder reactions, aldol reactions, radical cyclization, alkylation, and cascade reactions. 3n recent work, we reported the first use of a thiol-mediated acyl radical cyclization to construct the isotwistane skeleton from bicyclo[2.2.2]octenol 1 (Scheme 1), 4,5 which resulted in intermediate 2 needed for the biomimetic total synthesis of the palhinines.Our original plan was to complete the synthesis via isotwistane intermediate 4; however, when bicyclo[2.2.2]octanone 3 was reacted under thiol-mediated radical conditions, unexpected rearrangement product 5 was obtained instead of the desired cyclized product 4.A plausible rearrangement mechanism to account for this result entails the formation of cyclopropoxy intermediate B through a cascade 6-endo-trig/3-endo-trig cyclization from acyl radical A, followed by ring opening and hydrogen-atom transfer (HAT) to yield rearrangement isotwistane product 5. 6 Although the rearrangement product 5 was unexpected, the locations of its oxygen containing functional groups are consistent with those of the isotwistane skeleton of isopalhinine A, which inspired us to develop a more atomeconomic synthetic strategy toward the palhinines via the possible rearrangement intermediate 6 (Scheme 2).Accordingly, we synthesized bicyclo[2.2.2]octenone 7, hoping to then obtain rearranged product 6 by an acyl radical reaction.However, when bicyclo[2.2.2]octenones 7a and 7b were treated with tC 12 H 25 SH/AIBN, the direct cyclized products 8a and 8b 7 were obtained through 5-exo-trig cyclization instead.Modifying the functional groups on C3 or C6 did not affect the outcome of the reaction.Previous synthetic studies of the palhinines 8 concluded that the twisted structure of the isotwistane moiety may strain the 9-membered fused ring at its bridgeheads (C4 and C12).
In addition, a synthesis of a tricyclo[m.2.2.0]skeleton 9 via intramolecular Diels−Alder reactions of masked orthobenzoquinones (MOBs) 10 by Liao also reported the effect of ring strain on the yields of tricyclic products, with the yields of 7-membered ring products being significantly lower than those incorporating 6-or 5-membered fused rings.We hypothesized that a smaller 6-and 5-membered fused ring would be less strained at the bridgeheads than a 9-membered ring, and therefore attempted to reduce the fused ring size of the radical precursor to promote the formation of rearrangement product.
To investigate this hypothesis, 8-, 7-, 6-, and 5-membered fused ring precursors 9a−d were synthesized (Scheme 3) and subjected to the rearrangement reaction conditions.The 8-, 7-, and 6-membered precursors 9a−c were obtained from 3,4dimethoxylbenzaldehyde 10, 11 to which carbon chains of various lengths were appended using the Wittig reaction to give a selection of alkene products, which were hydrogenated to give carboxylic acids 11a−c.Friedel−Crafts cyclization reactions of 11a−c yielded the requisite 8-, 7-, and 6membered fused rings 12a−c, which underwent demethylation and Clemmensen reduction to yield the desired 2-methoxyphenols 13a−c.The 5-membered precursor 9d was obtained from 5-indanol 14, 12  Acyl radical precursors 9a−d were individually subjected to the thiol-mediated acyl radical reaction using tBuSH/AIBN; the results are presented in Table 1.As expected, reaction of the eight-membered fused ring 9a under these conditions yielded the cyclized product 17a in 54% yield, without any of the corresponding rearrangement product 18a (entry 1).For the seven-membered fused ring 9b, the desired rearrangement product 18b was obtained in 10% yield (entry 2), with the major products of this reaction being a mixture of the cyclization products 17b and 17b′, a related alkene side product, in a ratio of 4:3 and a total yield of 60%.For the sixmembered fused ring 9c, the corresponding rearrangement product 18c was obtained as a single product in 69% yield (entry 3), the structure of which was confirmed by X-ray  The strucure of products were confirmed by NMR spectra, 18c was fur-ther confirmed by X-ray crystallography.b The yield is isolated yield.c A mixture with alkene side product 17b′, the ratio 4:3 (17b:17b′) was determined by 1H-NMR spectrum.
crystallography. 13Surprisingly, the five-membered fused ring precursor 9d did not yield predicted rearrangement product 18d, but instead produced cyclization product 17d as a single product in 53% yield.
To further confirm the effect of ring strain on this cyclization/rearrangement process, we synthesized compound 19 (Scheme 4) bearing two methyl groups in place of the fused ring on the bridged alkene to eliminate its influence on the reaction.Both rearrangement (20) and cyclization (21) products were obtained in a ratio of approximately 1:1.This result is further evidence that the ring strain of the fused ring governs the outcome of the thiol-mediated acyl radical reaction of a bicyclo[2.2.2]octenone skeleton to give either a rearranged or cyclized major product.
Density functional theory (DFT) calculations were conducted to elucidate the reaction mechanism and rationalize the experimentally observed product selectivity.For reactant 19 lacking a fused ring, two possible pathways were identified, each of which leads to a different product (Scheme 5).Intermediate 19C is first formed by hydrogen abstraction of 19 using tBuSH/AIBN.We assumed that this step is facile and spontaneous, and therefore, its energetics were not computed.In the next step, the newly formed acyl radical couples with one carbon of the bridged alkene to form a C−C bond (19C → 19D).The reaction bifurcates at 19D.In the cyclization pathway (depicted in red), 19D abstracts a hydrogen from tBuSH to form cyclized product 20.In the rearrangement pathway, 19D forms a twistane intermediated 19E via 3membered ring rearrangement with the ketone on the 5membered ring, followed by another 3-membered ring rearrangement with the other ketone and hydrogen abstraction to form the rearranged product 21 (19D → 19E → 19F → 21, colored in blue).
Starting from 19D, where the two pathways branch, we found that the rate-determining step (RDS) for the cyclization pathway is 19D → 20, and 19D → 19E for the rearrangement pathway.It should be noted that the radical intermediate 19E can be generated directly from 19C and undergo a 6-endo-trig cyclization (19C → 19E).However, this pathway has a kinetic barrier that is 2.2 kcal/mol higher than that for 5-exo-trig cyclization (19C → 19D), suggesting it to be disfavored.
Gibbs free energy surfaces were calculated for 9a−d (see Supporting Information); the RDSs for the two pathways are the same as 19.The kinetic barriers of the RDS and reaction free energies for both pathways are summarized in Table 2 and compared to the observed product selectivity.Based on the kinetics, one would predict 9c and 9d to lead to the cyclization product, 9a and 9b to the rearrangement product, and 19 to both.This prediction is inconsistent with our previous  experimental observations.In contrast, a prediction based on thermodynamics (9a and 9d lead to the cyclization product, 9c to the rearrangement product, 19 and 9b to both products) is more consistent with our experiments, and suggests that, under the reaction conditions, all the kinetic barriers can be surmounted.Knowing that product selectivity is under thermodynamic control, the selectivity of the reaction can be predicted by comparing the relative energy levels of two potential products.
To further understand why starting material 9c led to the rearrangement product, a hypothetical homodesmotic reaction, as proposed by previous studies, 14 was used to calculate the ring strain energy.Products 20 and 21, both lacking fused rings, were selected as reference products due to their lack of ring strain and comparable thermodynamic stabilities.A key requirement for the homodesmotic reaction is that equal numbers of C, CH, CH 2 , and CH 3 components are on both sides of the equation.Therefore, in the calculation, the fused ring product 17 or 18 was reacted with ethane, removing the CH 2 component from the fused ring, and yielding propane and the fused ring free product 20 or 21 (Table 3).Taking cyclization product 17c as an example (where n equals one), four targeted CH 2 units (orange) were in the fused ring of 17c, while four CH 2 units were also present in four units of propane; ten CH 3 units (green) were in 5 units of ethane, while two CH 3 units were in 20, and eight CH 3 units were in propane.The negative values of the reaction enthalpies (−ΔH) in these homodesmotic reactions were used to estimate the ring strain energies.The calculated ring strain energies explain why six-membered fused ring entry 9c favored rearrangement product 18c.This preference can be attributed to the minimal ring strain, likely resulting from the chair conformation of the six-membered fused ring. 15n summary, we have demonstrated the effect of ring strain on an acyl radical reaction of a bicyclo[2.2.2]octenone skeleton leading to the formation of cyclized or rearranged isotwistane products.Based on these results and those of DFT calculations, the proposed rearrangement mechanism involves a two-step 3-membered ring rearrangement (D → E → F) proceeding via twistane intermediate E. Currently, we are exploring these pathways pursuant to a total synthesis of isopalhinine A and related natural products based on isotwistane or similar skeletons.We are also investigating the effect of other substituents on the outcome of this acyl radical reaction; the results will be published in due course.The general formula for calculating ring strain energy is based on a homodesmotic reaction.The orange and green colors indicate the CH 2 and CH 3 units, respectively, which remain equal on both sides of the equation.

■ ASSOCIATED CONTENT
which led to desired 2-methoxyphenol 13d by one-pot bromination and an Ullman-type coupling.As depicted in Scheme 3b, MOB intermediates 15a−d were obtained by oxidative dearomatization of 13a−d followed by a Diels−Alder reaction with acrolein to yield bicyclo[2.2.2]octenones 16a−d.Finally, acyl radical precursors 12a−d were obtained by the one-carbon elongation of 16a−d via Wittig reactions and acidic hydrolysis.

Scheme 4 .a
Scheme 4. Model Study of Thiol-Mediated Acyl Radical Reaction without Fused Ring

Table 1 .
Scheme 2. Our Failure To Obtain Rearrangement Product 6 from Bicyclo[2.2.2]octenone 7 Scheme 3. Syntheses of Radical Rearrangement Reaction Precursors 9a−d with Reducing Size of Fused Ring Results of Thiol-Mediated Acyl Radical Reaction with Different Ring Size of Fused Rings

Table 2 .
Calculated Overall ΔG ‡ and ΔG for 19 and 9a−d for Both Pathways

Table 3 .
Calculated Strain Energies a