Deoxygenative Olefin Insertion of Cyclic Alcohols Promoted by Sulfoxide Cation Radicals

Abstract

Herein we report a novel strategy to expand aliphatic cyclic alcohols via deoxygenative olefin insertion. Acridinium-catalyzed formation of sulfoxide radicals selectively converts unfunctionalized alcohols to alkoxy radicals, which undergo selective β-scission of the weakest adjacent C–C bond. The resulting alkyl radical undergoes a Giese addition with a vinyl phosphonate acceptor. Subsequent treatment of this ketophosphonate with mild base allows for a highly efficient Horner–Wadsworth–Emmons (HWE) reaction to proceed, generating the two-carbon-expanded product. The products notably possess an internal double bond, which is unique with respect to ring expansion methodologies. Further, this approach does not rely on preactivated starting alcohols, is applicable to the formation of 5–7-membered cycloalkenes, and proceeds with a broad array of functional group tolerance in moderate to excellent yields.

Graphical Abstract

INTRODUCTION

Aliphatic ring systems are key structural motifs in natural products, pharmaceuticals, polymers, and ligands due to their structural rigidity, spatial orientation, and stereochemical scaffolds.^1–4 These geometric and conformational properties have led both chemists and nature to heavily explore substitution along this cyclic core as a means of generating predictable, yet complex three-dimensional frameworks. Accordingly, the development of methods that enable the expansion of readily accessible smaller rings to their larger counterparts provide an appealing strategy for their synthesis.⁵ This is especially noteworthy with respect to 7-membered rings, which are significantly more difficult to synthesize than 5 and 6-membered rings due to an increase in ring strain.⁶ Through such expansion reactions, one may modulate the core structure of a given compound.⁷

Fundamental examples of directly expanding an aliphatic ring system containing an alcohol/ketone by n + 1 include the Baeyer–Villiger oxidation, Schmidt reaction, Beckman rearrangement, Büchner–Curtius–Schlotterbeck reaction, and the pinacol rearrangement.^8,9 However, significantly fewer examples exist capable of n + 2 ring expansions. The De Mayo reaction is a traditional approach to accomplish this ring expansion (Scheme 1A).^10–12 Many others have also contributed by expanding both the nucleophile and electrophile scope, but the requirement for beginning with 1,3-dicarbonyl-containing cycloalkanones remains.^13–15 Similar, but mechanistically distinct work using cyclohexyne has also been developed by Carreira for the n + 2 ring expansion of ketones.¹⁶ Dong described a rhodium-catalyzed direct insertion of ethylene into 1-indanones to provide access to benzocycloheptenones via a “cut-and-sew” strategy (Scheme 1B).^17,18 Zuo reported a cerium-catalyzed expansion of cyclobutanols and cyclopentanols to bridged lactones via alkoxy radical generation, finally converting the intermediates to products with the addition of strong acid and heating (Scheme 1C).^19,20 Notably, the cyclopentanol scope of this method was largely restricted to 2-indanols, as the formation of the corresponding seven-membered bridged lactone via the intramolecular aldol step proved entropically challenging for many substrates. Further, Knowles developed a proton-coupled electron transfer (PCET)-based approach from allylic alcohols capable of furnishing either the n + 1 or n + 2 product depending on the substitution of the starting material (Scheme 1D).²¹ Since then, many ring-expansion strategies relying on intramolecular radical addition to π-systems have been developed.^22–29

Scheme 1.

(A–D) Existing Ring Expansion Methodologies;(E) This Work

Despite these advances, the substrates required for direct n + 2 ring expansion nearly always necessitate preactivation or prefunctionalization, which restricts the potential distribution of products that may be generated (e.g., 1,3-dicarbonyls, benzocycloalkanes, allylic alcohols, etc.). Similarly, approaches that do not require traditional preactivation are commonly restricted to specific starting materials that feature a high degree of ring strain (e.g., cyclopropanes and cyclobutanes, >20 kcal/mol) or require Thorpe-Ingold effects for the resulting cyclization to proceed efficiently. In addition to requiring rare earth transition metals, many of these strategies also use unattractive conditions for their expansions in the forms of pressurized systems, high temperatures, long reaction times, strong acids, and starting materials derived from organometallics (aluminum hydrides, Grignards, and lithiates, especially).

Finally, the vast majority of these approaches commonly generate ring systems that maintain a carbonyl in the expanded ring: examples of this are ubiquitous in both the traditional and modern strategies for n + 1 and n + 2 ring expansions. While ketones remain a highly versatile functional group, the discovery and development of complementary methods that may provide access to disparate yet equally useful motifs are necessary to improve a chemist’s synthetic toolbox.^30,31

In the past decade, recent photochemical methods have been developed to generate alkoxy radicals and harness their innate reactivity with respect to β-scission reactions. Through this process, a strong carbonyl bond may be formed via the cleavage of an adjacent C–C bond, leaving a carbon-centered radical in the process.³² Work by Zuo and others focused on ligand-to-metal charge transfer (LMCT) strategies with cerium and titanium Lewis acid catalysts have utilized this approach to perform ring opening as well and stereochemical ring editing reactions of unactivated alcohols.^33–37 PCET strategies centered around oxidizing iridium photoredox catalysts have also been developed.³⁸ When performed on an appropriately reactive cyclic alcohol, the result of these strategies is effectively the same: the generated alkoxy radical cleaves its weakest adjacent C–C bond to form a carbonyl and alkyl radical, which may be trapped downstream. Despite these advances, strategies leveraging the predictable reactivity of alkoxy radicals are of increasing synthetic importance; therefore, our group endeavored to develop a strategy toward their generation.

In 2024, we disclosed a strategy utilizing sulfoxide cation radicals to generate alkoxy radicals.³⁹ These versatile intermediates were demonstrated to undergo cyclizations, 1,5-hydrogen atom transfer (1,5-HAT), and β-scission reactions; the resulting alkyl radicals were then capable of being captured by several Giese acceptors to yield the functionalized products. We became interested in utilizing this strategy as part of a sequential ring-opening and ring-closing manifold to enable multi-atom ring expansions. While previous reports using this strategy perform a subsequent radical cyclization to generate the ring-expanded product, we were instead focused on using an anionic Horner-Wadsworth-Emmons (HWE) Reaction (Scheme 1E).^40–44 This approach would come with the added benefits of not requiring prefunctionalized starting alcohols and generating products with a highly polarized internal olefin, allowing for broader functional group tolerance in the starting materials while providing access to products in new chemical space.

RESULTS AND DISCUSSION

We began our investigations utilizing a vinyl phosphonate acceptor, which we hypothesized would be compatible with our lab’s alkoxy radical generation protocol due to its highly electron-deficient nature (Table 1). Utilizing the adopted conditions from our lab’s previous report (Table 1, entry 1), we were pleased to see radical addition into the acceptor but disappointed that no cyclized product was obtained. The basis for this transformation was a hypothesis that the anion generated by the reduction of the intermediate radical species would perform the HWE reaction, as has been documented in the literature with Wittig-type reagents in photoredox reactions.^45,46 We later found that using the alcohol as the limiting reagent could actually produce this intermediate in excellent yield, so long as a different pyridine base (which proved vital for reactivity) and higher loadings of sulfoxide were used (Table 1, entry 5). This is a significant improvement on our previously reported conditions, which required the precious starting alcohol to be used in 5 equiv relative to the acceptor and reacted for 48 h.

Table 1.

Reaction Optimization of Olefin Insertion^a

entry	deviation from standard conditions (above)	intermediate (%)	product (%)
1	5:1 alcohol:acceptor, 1.0 equiv sulfoxide, 0.5 equiv pyridine, 48 h, no LiCl/DBU cyclization	22	<5
2	5:1 alcohol:acceptor, 1.0 equiv sulfoxide, 0.5 equiv pyridine, no LiCl/DBU cyclization	20	<5
3	1:3 alcohol:acceptor, 1.0 equiv sulfoxide, 0.5 equiv pyridine, no LiCl/DBU cyclization	<5	<5
4	1:3 alcohol:acceptor, 1.0 equiv sulfoxide, no LiCl/DBU cyclization	67	<5
5	1:3 alcohol:acceptor, no LiCl/DBU cyclization	87	<5
6	1:3 alcohol:acceptor	<5	59
7	no deviation	<5	68 (64)
8	PhCF3, DCM, or DCE	<5	59–64
9	LiCl and DIPEA as base for HWE (21 h)	<5	49
10	no sulfoxide, 2,6-lutidine, or acridinium	<5	<5
11	no irradiation	<5	<5

^aYields determined via ¹H NMR using HMDSO as an internal standard. Alcohol is 3:1 trans:cis. Yields in parentheses indicate isolated yields. PhCF₃ = trifluorotoluene, DCM = dichloromethane, DCE = 1,2-dichloroethane, MeCN = acetonitrile, LiCl = lithium chloride, DBU = 1,8-Diazabicyclo[5.4.0]undec-7-ene, DIPEA = diisopropylethylamine.

Directly adding LiCl and DBU to the reaction following irradiation and stirring for 4 h at room temperature fortunately provided the cycloheptene product in 59% yield with trace aldehyde remaining (Table 1, entry 6). Lowering the acceptor equivalents presumably reduced side reactions and furnished the product in a 64% isolated yield (Table 1, entry 7). The reaction sequence proved robust, as it was capable of generating the cyclized product in several solvents, even those not commonly reported for an HWE reaction (Table 1, entry 8). DIPEA, a milder alternative to DBU, was also capable of furnishing the product, given an extended reaction time (Table 1, entry 9). Removal of any component of the photoredox reaction resulted in no formation of the intermediate (Table 1, entries 10 and 11). Attempts at utilizing either catalytic cerium- or iridium-based strategies for the alkoxy-radical generation procedure resulted in diminished levels of intermediate formation, suggesting that the sulfoxide cation radical-based approach is particularly well-suited to accomplish this transformation (see Supporting Information). Additionally, catalytic loadings of sulfoxide can be used in lieu of stoichiometric quantities with increased reaction times (see Supporting Information).

With the optimized conditions in hand, the scope of the transformation was next explored with respect to the ring expansion of cyclopentanols to cycloheptenes (Scheme 2). We found that primary, secondary, and tertiary radicals could be selectively formed, captured, and subsequently cyclized in moderate to good yields (1–3). Cyclopentanols containing a ketal (4) and acetate (5) at the 2-position also demonstrates the compatibility with α-oxygen centered radicals. Protected primary amines in the form of secondary amides were additionally well tolerated in the transformation (6). X-ray analysis of 6 revealed that the cycloheptene core favors a chairlike conformation. This lends credence to our suspicions that the HWE requires the product not to possess a high degree of strain energy for intramolecular cyclization. In support of this hypothesis, attempts at cyclizing intermediates derived from cyclohexanols resulted in only trace yield of the cyclooctene products which maintain a higher strain energy than cycloheptenes (see Supporting Information).^47,48 In addition to this, we also found tertiary alcohols (1-methyl substituted specifically) easily scissioned to form the intermediate; however, the resulting HWE proceeded in poor yields, presumably due to the decreased electrophilicity of the ketone (see Supporting Information).

Scheme 2. Scope of Olefin Insertion in Cyclopentanols.

^aScission reaction performed with 5 mol % Mes-Acr-BF₄, 1.0 equiv collidine, and 1.2 equiv acceptor. ^bIsolated yield of the gram-scale batch reaction. ^cCyclized with 3.0 equiv NaH for 20 h. ^dCyclized for 20 h under standard conditions.

A common strategy for synthesizing medium-sized rings involves ring-closing metathesis utilizing the Hoveyda-Grubbs generation II catalyst with two terminal olefins.^49–51 In addition to requiring multiple synthetic steps to access the starting materials, this approach would likely not be applicable toward the formation of products with terminal alkenes such as 7 as the precursor would possess too many reactive motifs. In a similar vein, alkynes were additionally tolerated in good yield (8). Both electron-poor and -rich arenes were also tolerated in a pyrazole-based heterocycle (9) and PMP group (10). Accounting for the latter’s high yield, quantitative intermediate formation was observed with only collidine and Mes-Acr-BF₄ prior to cyclization via a PCET-based mechanism analogous to prior reports.^52,53 Ketoamide-containing product 11 was capable of being synthesized in good yield without any off-target HWE reactivity occurring. This result is in line with expectations, as the aldehyde is not only a better electrophile but also an HWE olefination on the ketone would form a thermodynamically unfavorable cyclononene. Multiple nitrogen-containing functional groups in one compound were also tolerated, as demonstrated with 12, containing both an amide and a pyridine heterocycle. Fluorine, chlorine, and bromine (potential cross-coupling partners) were all tolerated in moderate to good yields, as displayed in the benzamide-containing compounds 13 and 14.

Substituted tetrahydrooxepines (15 & 16) were also generated from the corresponding easily accessible 3-hydroxytetrahydrofurans. These compounds can be further derivatized to access medicinally relevant oxepanes.^54,55 Disappointingly, attempts at synthesizing the corresponding nitrogen-containing azepane resulted in either a failure to generate the intermediate or a failure to cyclize (see Supporting Information).

After demonstrating a broad array of functional groups compatible with this two-step, one-pot protocol, we sought to investigate more complex cyclopentanols. Cycloheptene 17 could be synthesized in moderate yield from a loxoprofen-derived starting material (originally 4:4:1:1 d.r.) as a 1:1 mixture of diastereomers.

Additionally, cyclopentanols derived from natural products such as methyl-dihydrojasmonate can be utilized to forge cycloheptenes, which afforded 18 (50% yield, 1:1 d.r.) and features substitution at the 3-position of the cyclopentanol. This transformation was also shown to be amenable to scale-up, with 18 synthesized using a batch protocol on a preparative scale with effectively no loss in yield. A uniquely adorned cycloheptene core can be generated via substitution at the 4-position of the cyclopentanol, as demonstrated in example 19, in good yields. Sodium hydride was required to cyclize this intermediate, presumably due to the presence of esters, which can occupy the attention of lithium and prevent DBU-mediated deprotonation. Additionally, a reduced derivative of trans-androsterone provided the ring-expanded product 20 in 42% yield. The positioning of the olefin in the 6,6,6,7 ring system is notable as it allows for easy diversification along the cycloheptene core to access unnatural terpenoid derivatives, which are difficult to synthesize by other routes.⁵⁶

After exploring the ring expansion of cyclopentanols, we turned to the synthesis of cyclopentenes and cyclohexenes from cyclopropyl alcohols and cyclobutanols, respectively (Scheme 3). To this end, cyclopentene 21 and cyclohexene 22 were formed in good yields following our standard reaction conditions. Cyclohexenoate 23 represents an example of a spirocyclic cyclobutanol undergoing ring expansion while tolerating a tosyl-protected amine. Similarly, four-membered heterocycles also underwent ring expansion, with dihydropyran 24 obtained in good yield from commercially available oxetane-3-ol. A Boc-protected azetidine variant could also be expanded to 25 in 44% yield.

Scheme 3. Expansion of Strained Rings and Insertion of Various Olefins^e.

^aCyclized for 20 h under standard conditions. ^bCyclobutanol was used as the substrate. ^c3.0 equiv acceptor was used. ^dCyclopropanol was used as the substrate. ^eCyclized with 3.0 equiv NaH for 20 h.

We next examined whether ring expansion could take place to form a tetrasubstituted cyclohexene via intramolecular cyclization on a ketone. Indeed, using 1-methylcyclobutanol and simply running the standard HWE reaction for 20 h afforded cyclohexenoate 26 in 63% yield. The increased rate of nucleophilic attack for a 6-exo-tet vs 7-exo-tet as well as the improved thermodynamic stability of a cyclohexene vs cycloheptene could explain this result.⁵⁷ Additional substitution on position 3 of the cyclobutanol starting material formed highly substituted cyclohexene 27 in nearly quantitative yield. Groups larger than methyl could also be included off the olefin (n-Bu) alongside a benzyl ether in a distal position of 28 in synthetically useful yield.

Up until this point, all products featured an acrylate as the olefin insertion motif. We synthesized several Giese acceptors with various substitution patterns to further expand the alkene chemical space accessible by this transformation. Unsaturated amide 29 was synthesized in excellent yield by utilizing a secondary amide-based trap. Enone 30 was also accessed in a high yield with a ketone-based acceptor. Cyclohexenyl nitrile 31 was formed with no olefin transposition to the tetrasubstituted thermodynamically favorable alkene, likely due to the mild reaction conditions of the cyclization. Finally, cyclohexenyl phosphonates can also be synthesized (32) from the corresponding bisphosphonate Giese acceptor.

Given our lab’s previous report on the generation of alkoxy radicals mediated by sulfoxide cation radicals, we propose the following mechanism, demonstrated with the formation of 1 (Figure 1).³⁹ Excitation of the acridinium photoredox catalyst with visible light (456 nm) generates the excited state Xyl-F-Acr-BF₄* complex (E^red_1/2* = +2.13 V vs SCE). A single-electron transfer event then occurs with diphenyl sulfoxide (E_p/2* = +2.06 V vs SCE) to produce the sulfoxide cation radical intermediate. It is worth noting that this oxidation is nearly thermoneutral, meaning that the electron transfer to generate the intermediate is relatively slow. During optimization, we found that increasing the sulfoxide equivalents can greatly increase the reaction rate—doing so minimizes the overall reaction time and allows for increased throughput. Additionally, the reaction can still proceed in high yields with a catalytic loading of sulfoxide, given an increase in the reaction time (see Supporting Information). Continuing, nucleophilic addition of the alcohol, facilitated by deprotonation with base (2,6-lutidine) into the sulfoxide cation radical, generates sulfurane I. Two β-scissions first yield alkoxy radical II and subsequently the ring-opened aldehyde-containing alkyl radical III. Giese addition into the polarity-matched electron-deficient vinyl phosphonate acceptor gives α-phosphonyl radical IV. Reduction from Xyl-F-Acr• (E_ox* = −0.54 V vs SCE) closes the catalytic cycle in a manner similar to past reports from our laboratory, and protonation yields the isolable pronucleophile V.^58–61 Upon completion of the photoredox reaction, simple addition of LiCl and DBU furnishes ring-expanded cycloheptene product 1 via a Horner-Wadsworth-Emmons (HWE) reaction.

Figure 1.

Proposed Mechanistic Pathway.

CONCLUSION

In conclusion, we have developed a deoxygenative procedure to insert olefins into unfunctionalized cyclic alcohols. 3, 4, and 5-membered rings were all expanded while demonstrating a broad array of substitution patterns and functional group tolerance. This approach was made possible by leveraging the β-scission reactivity of alkoxy radicals generated via an acridinium-enabled sulfoxide cation radical protocol. The resulting ring-opened intermediates undergo an intramolecular HWE reaction to furnish the expanded products. Future directions involve exchanging the Giese acceptor for other motifs to allow for the development of additional novel ring expansion methodologies.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c23176.

Experimental procedures; optimization and controls; materials and methods; HRMS data; supporting ¹H and ¹³C NMR spectra (PDF)