Synthesis of Vicinal Quaternary All-Carbon Centers via Acid-catalyzed Cycloisomerization of Neopentylic Epoxides.

We report our studies on the development of a catalytic cycloisomerization of 2,2-disubstituted neopentylic epoxides to produce highly substituted tetralins and chromanes. Termination of the sequence occurs via Friedel-Crafts-type alkylation of the remote (hetero)arene linker. The transformation is efficiently promoted by sulfuric acid and proceeds best in 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) as the solvent. Variation of the substitution pattern provided detailed insights into the migration tendencies and revealed a competing disproportionation pathway of dihydronaphthalenes.

Vicinal quaternary carbon centers are present in many bioactive natural products (e.g., salimabromide (1),[1] lingzhiol (2),[2] calycanthine (3),[3] communesin F (4),[4] and koumine (5)[5]) and pharmaceuticals such as buprenorphine (6)[6] (Scheme A). The presence of these structural units was reported to increase the structural rigidity allowing for tighter binding to their molecular targets in many cases and greater selectivity than with more flexible congeners.[7] In nature, all-carbon quaternary centers are, for instance, accessible via reactions that proceed via carbocation intermediates.[8,9] For their construction in the chemical laboratory, a well-assorted toolbox has been established in the past.[9,10] However, synthetic challenges remain, as multistep procedures that are accompanied by low-yielding transformations are often required.

Scheme 1

Occurrence of Vicinal All-Carbon Quaternary Centers in Bioactive Molecules and Concept of this Work

In the context of the synthesis of salimabromide (1), we were investigating methods to efficiently construct the fully substituted tetrahydronaphthalene core.[11] We found that ring formation and installation of the two crucial vicinal quaternary carbon centers were possible in a single step by means of a powerful cycloisomerization reaction of a 2,2-disubstituted neopentylic epoxide. This chemistry was inspired by the seminal reports by Bogert[12] and Cook[13] in 1933 (Scheme B). In this work, a tandem hydride migration/Friedel–Crafts-type cyclization of tertiary alcohol 7 enabled the synthesis of octahydrophenanthrene system 8. In 2010, Khalaf extended the rearrangement–cyclization cascade by resorting to acyclic tertiary alcohols such as 9 to enable installation of two vicinal gem-dimethyl groups.[14]

Unfortunately, both of these reports were strictly limited to a few unfunctionalized hydrocarbon frameworks. Herein we disclose the synthesis of vicinal all-carbon quaternary centers by the consecutive 1,2-rearrangement/cyclization of 2,2-disubstituted neopentylic epoxides under mild conditions (Scheme C). Selective migration of various alkyl residues was achieved by exploiting ring strain, carbon–carbon bond strengths, and carbocation stabilities. This allowed for the synthesis of a library of polyfunctionalized tetralin and chromane systems.

For the initial optimization of the reaction conditions, we employed readily available electron-rich arene 11a (Scheme A). We were pleased to find that the cycloisomerization proceeded most efficiently in 1,1,1,3,3,3-hexafluoroisopropanol (HFIP)[15] at 0 °C with sulfuric acid (10 mol %) as catalyst, affording tetralin 12a in 83% yield within 15 min. Alternative solvents and Brønsted or Lewis acids were found to be inferior and led to significantly reduced yields (entries 2–7).[16] Higher temperatures (23 °C, entry 9; 58 °C, entry 8) were less effective for this transformation, leading to complex mixtures of uncyclized byproducts.

Scheme 2

Optimization and Scope of the Cycloisomerization

1H NMR yields with 2,3,5,6-tetrachloronitrobenzene as the internal standard are shown, with isolated yields in parentheses. n.o. = not observed.

With the optimized conditions in hand, we investigated the scope of the cycloisomerization in more detail (Scheme B). As a first parameter, we studied different aromatic residues as the nucleophilic component for the Friedel–Crafts termination step. Activating methyl and tert-butyl substituents provided yields of up to 83% (12c and 12d). Methoxy-substituted tetralins were obtained in virtually identical yields as for unsubstituted tetralin 12b, with only a little influence of the substitution pattern (69% for 12e and 70% for 12f). It is noteworthy that 12e was previously synthesized by the same cascade under nonoptimized conditions in only 43% yield.[11] Remarkably, a fully methylated pyrogallol-derived epoxide formed the corresponding tetralin 12h in only 57% yield, while benzodioxole derivative 12g was isolated in 82%. We believe that this results from the trajectory of the approaching arene, which leads to severe steric interaction between the outer methoxy groups and the tertiary carbocation unit. As expected, substrates with deactivating substituents delivered the corresponding tetralins in only low yields or completely shut down the reaction (see Scheme D, limitations). Fluorinated tetralins 12i and 12j were formed in 35 and 41% yield, respectively. Pinacol boronate 12k, which is a valuable building block for further derivatizations via Suzuki–Miyaura cross-coupling reactions, was formed in 29% yield. Electron-rich, nonbasic heterocycles also proved to be compatible with the reaction conditions. For a thiophene substrate, efficient alkylation took place to afford the annealed 6/5-system 12l in 77% yield. Furans 12m and 12n were formed in lower yields under the reaction conditions, probably because of competing hydrolysis or polymerization.[17] We were pleased to see that the methodology is not limited only to aromatic nucleophiles: oxane 12o was formed in 43% yield from the corresponding primary alcohol. Interestingly, even for this relatively small nucleophile no oxolane formation was observed. This underpinned our assumption that the formation of a less-strained six-membered ring must be one of the major driving forces for the cascade (vide infra). This hypothesis was also experimentally supported by product 12p, which was formed from a distal epoxide via a formally inverse methyl migration along the alkyl chain. The non-rearranged 7/6 system was not observed under these conditions.

Having investigated the conversion of a panel of variously substituted (hetero)arenes to afford tetralins, we proceeded to vary the carbon chain connecting the epoxide and (hetero)arene (Scheme C). By replacement of the ethylene linker with an oxymethylene unit, we envisioned being able to access heterocyclic chromane systems. We were pleased to see that a series of these readily available phenyl ethers delivered highly substituted chromanes 13a–l in medium to good yields. For all of the substrates investigated, a competing hydride shift to form a stabilized phenoxycarbocation ion was not observed. Electron-rich phenyl and naphthyl ethers underwent the cascade reaction in up to 76% yield. Similarly substituted chromanes were generated in only slightly reduced yields compared to the corresponding tetralin analogues. An exception was methoxy derivative 13d, which was formed in only 33% yield together with a 7% yield of its ortho regioisomer 13e. Interestingly, we observed an unexpected effect of the substitution pattern of methoxyphenols, affording the highest yield (58%) for the ortho-substituted derivative 13f. The p-fluoro- and p-chlorophenyl ethers delivered the corresponding chromanes 13i and 13j in 73% and 68% yield, respectively. For substrates carrying a (boronic) ester (CO2Et or Bpin), only small amounts of the corresponding chromanes were isolated (23% yield for 13k and 39% yield for 13l). In these cases, the phenyl ether proved to be less stable, leading to the isolation of free phenols in substantial amounts. Electron-poor (hetero)arenes turned out to be incompatible not only for the formation of tetralins but also for chromane systems (Scheme D). The 2,3-substituted pyridine 12q was not observed even in the presence of excess acid and prolonged reaction times. A trifluoromethyl group (12r), amide (12s), or nitrile (13m) prevented the final cyclization and gave mostly complex mixtures of uncyclized elimination products. To our surprise, protected aniline derivative 13n was not accessible either, and only a complex mixture was obtained.

To better understand the reaction sequence, we studied the migration tendencies of different alkyl residues (Scheme ). For this purpose, we replaced the tert-butyl group attached to the epoxide with different aliphatic rings (ring size = 4, 5, 6). Highly strained methylcyclobutyl epoxide 14a exclusively underwent ring expansion/alkyl migration, affording cis-benzohydrindane 16a in 76% yield. The same behavior was observed for the formation of tetralin 16d (82%) and chromane 16e (87%). The less-strained cyclopentyl derivative 14b afforded both cis-benzodecalin 16b and spirane 15b as an inseparable mixture (9:1 ratio), still preferring the ring-expanded product 16b in 54% yield. Cyclohexyl derivative 14c exclusively formed spirane 15c (71%) with no detectable amount of the 7/6/6-expansion product 16c.

Scheme 3

Investigation of the Migration Tendencies of Substituents

In an effort to investigate the requirements for successful 1,2 migration, we further modified the tert-butyl group and replaced one of the three methyl groups with an allyl, prenyl, benzyl, or vinyl group. Epoxide 17a afforded allyl-migrated tetralin 18a in 38% yield as the major product. Preferential migration of the weaker allylic bond was also observed for the prenyl substrate 17b. For this particular case, we observed protonation of the remote double bond of 18b and subsequent spiroxane formation with the neopentylic alcohol (39% yield of 19). Similar migration was observed for the benzyl group, affording 18c in 36% yield. The competing methyl migration was also observed for this substrate, leading to the formation of 20 as a mixture of diastereomers (1.6:1 d.r.) in 19% yield. Careful analysis of the product mixture revealed the unusual 6/6/6/6-product 21 (hexahydrobenzo[c]phenanthrene) as the major product (41%). Despite the stronger sp2–sp3 bond (t-Bu–vinyl = 97.8 kcal mol–1 vs t-Bu–methyl = 87.5 kcal mol–1),[18] the migration of a vinyl group was also observed to afford tetralin 18d in 22% yield together with a complex product mixture. The eight-membered-ring product 22, originating from at least two alkyl migration steps, was isolated as the only byproduct in 7% yield.[19]

To investigate the requirements for successful migratory cycloisomerization, we varied the degree of substitution of the terminal alkyl carbon starting with a methyl group (23a, R1 = R2 = R3 = H). As expected, no cycloisomerization was observed for this epoxide, but disproportionated naphthalene 27a (47%) and tetralin 28a (49%) were obtained in nearly quantitative combined yield. The same results were observed for substrates carrying an ethyl (23b) or benzyl (23c) group. When an isopropyl group was present, naphthalene 27d (29%) and tetralin 28d (34%) still prevailed. However, a 1,2-hydride shift was also observed to afford cycloisomerized tetralin 24d in 28% yield. Diphenylmethyl derivative 23e afforded the corresponding products in similar yields. Surprisingly, in this case only phenyl migration with low diastereoselective control (23e, 36% yield, 1.9:1 d.r.) was observed. Epoxide 11a, which was used for the optimization, delivered disproportionated naphthalene 27f (8%) and tetralin 28f (9%) under the reaction conditions. For the methoxymethyl group in substrate 23g, we exclusively observed direct alkylation leading to a mixture of dihydronaphthalene 26g and its disproportionation products, the corresponding naphthalene 27g and tetralin 28g.[19] Fast disproportionation was observed not only for dihydronaphthalenes but also for cycloisomerized neopentylic thiol 30. The use of thiirane 29 directly gave a mixture of desulfurized tetralin 31 (41%) and the corresponding disulfide 32 (22%).[20] While the disproportionation of thiols to disulfides and hydrogen is a common reaction,[21] the formation of a hydrocarbon and a disulfide is unprecedented to the best of our knowledge.

Finally, we also screened a panel of chiral Lewis and Brønsted acid catalysts employing substrates 11a and 17d. Unfortunately, we did not observe any asymmetric induction (see the Supporting Information for screening).[22] It is noteworthy that for all of the substrates investigated, no five- or seven-membered-ring systems were observed.[23]

In conclusion, we have reported a powerful cycloisomerization reaction of 2,2-disubstituted neopentylic epoxides. The reaction does not require transition metal catalysts and proceeds under mild conditions in HFIP as the solvent. Variation of the terminating nucleophile enabled rapid access to functionalized chromanes and tetralins featuring vicinal all-carbon quaternary centers. The use of cycloalkyl moieties allowed for the formation of tricyclic ring systems in one step. Analysis of the byproducts revealed fast disproportionation of dihydronaphthalenes to form naphthalenes and tetralins.