Ynol Ethers as Ketene Equivalents in Rhodium-Catalyzed Intermolecular [5 + 2] Cycloaddition Reactions

Abstract

The previously unexplored metal-catalyzed [5 + 2] cycloadditions of vinylcyclopropanes (VCPs) and electron-rich alkynes (ynol ethers) have been found to provide a highly efficient, direct route to dioxygenated seven-membered rings, a common feature of numerous natural and non-natural targets and building blocks for synthesis. The reactions proceed in high yield at room temperature and tolerate a broad range of functionalities. Substituted VCPs were found to react with high regioselectivity.

Graphical abstract

New reactions, reagents, and catalysts change how we think about bond construction, thereby enabling new strategic choices for step economical and greener, if not ideal, syntheses.¹ As part of our studies on new cycloaddition reactions,^2,3 we previously reported a route to seven-membered rings involving the metal-catalyzed [5 + 2] cycloaddition of vinylcyclopropanes (VCPs) and π-components.⁴ Rhodium catalysts have proven to be the most general for this CC bond activation process, working thus far intramolecularly with alkynes, alkenes and allenes and intermolecularly with alkynes and activated allenes as 2C components.^2,3,5

To extend the reach of these [5 + 2] cycloaddition reactions and more generally other [m + n] processes, we have been exploring the use of π-component equivalents of otherwise inaccessible, difficult to use, or unsafe π-systems including allene⁶ and buta-1,2,3-triene^4c equivalents of gaseous allenes and cumulenes as well as tetramethyleneethane⁷ (TME) equivalents of the unstable and difficult to access TME. Here we report the use of ynol ethers (Scheme 1, left) as ketene (Scheme 1, right) equivalents in [5 + 2] cycloadditions with VCPs.⁸

Scheme 1.

Use of Ynol Ethers as Ketene Equivalents

While ketenes can be used as π-components in some metal-catalyzed cycloadditions,⁹ their electron-poor nature, propensity to dimerize, and incompatibility with a range of functionalities limits their utility.¹⁰ In contrast, ynol ethers are electron-rich and easily prepared by alkylation of the parent metal alkoxyacetylide.¹¹ However, their use in metal-catalyzed cycloadditions is largely unexplored and potentially problematic due to their reported “instability” in the presence of cationic rhodium complexes.^12,13 Beyond their mechanistic interest, the study of ynol ethers as 2C components in [5 + 2] cycloadditions is further motivated by the potential use of such a process in accessing diverse targets.^2,14 Numerous natural (estimated at >3000)¹⁵ and non-natural products, including many of research and therapeutic importance,¹⁶ incorporate functionalities derivable from cycloheptan-1,4-diones (CHDs).^17,18 Yet few methods exist for the direct construction of such systems.¹⁹ We have now found that the metal-catalyzed cycloaddition of ynol ethers and VCPs provides a solution to this problem.

To determine the suitability of ynol ethers as substrates²⁰ in [5 + 2] cycloadditions, 1-ethoxy-1-octyne (2c, R = n-hexyl) was chosen as a test reactant in an initial catalyst screening (for substrate syntheses, see Supporting Information (SI)).

We first tested [RhCl(CO)₂]₂ as a catalyst in the reaction of 2c at 25 °C with commercially available VCP 1a. Cycloadduct 3c did not form. Upon heating at 90 °C, the reaction gave 3c albeit in only 52% yield. A recently introduced cationic Rh(I) catalyst ([Rh(dnCOT)(MeCN)₂]SbF₆)^4a,b provided only complex mixtures. In contrast, [Rh(naph)(COD)]SbF₆, another cationic Rh catalyst,²¹ gave promising initial results (Table 1, entry 1:60% of 3c), working even at 25 °C in 2,2,2-trifluorethanol (TFE), and was thus selected for further study.

Table 1.

Optimization Studies

entry	solvent	catalyst loading [mol %]	ynol ether [equiv]	t [h]	yield [%]
1	TFE	3	1.1	8	60
2	TFE	3	3.0	12	35
3	TFE	3	1.1	4.5	0a,b
4	TFE	5	1.1	4	74
5	TFE	5	1.1c	2	91
6	TFE	0	1.1	4	0d

^a[Rh] and 2c were stirred in TFE for 2 h prior to the reaction.

^b1a was recovered.

^c2c was added dropwise over the course of 2 h.

^dReaction was carried out at 80 °C.

Interestingly, when excess ynol ether 2c (3.0 equiv) was used to increase the yield, cycloadduct 3c was obtained but in only 35% yield, suggesting that the ynol ether inhibits catalysis (Table 1, entry 2). To test this point, the catalyst was stirred with ynol ether 2c for 2 h after which VCP 1a was added (entry 3). No cycloadduct was formed and only starting materials were isolated. To overcome this substrate inhibition problem, the catalyst loading was increased (5 mol %) and the amount of the ynol ether was decreased (1.1 equiv, entry 4). An improved yield (74%) was obtained. Finally, to further minimize the inhibitory effect of the ynol ether, 2c was added dropwise over 2 h. Under these conditions, the cycloadduct was formed in excellent yield (91%, entry 5). No reaction was observed in the absence of catalyst, even when the reaction was heated for 4 h (entry 6).

Using the above conditions, a broad range of ynol ethers yielded CHDs in good to excellent yields (Scheme 2). Terminal alkyne 2a (EtOCCH) and TMS-analogue 2b (EtOCCTMS), equivalents of ketene itself, reacted efficiently, both giving dione 3a after workup. Alkyl-substituted (2c–2g, 2i–2j) and aryl-substituted ynol ethers (2k–2q) were also effective substrates. Halogen containing substrates (2e, 2f) reacted efficiently along with terminal alkene 2g (84% yield). Of mechanistic interest, trisubstituted alkene 2h gave a complex mixture, potentially due to catalyst deactivation by chelative coordination. Supporting this hypothesis, the otherwise efficient reaction of 2d with VCP 1a, in the presence of 2h, yielded no cycloadduct 3d. Benzyl substituted ynol ethers (2i and 2j) also worked moderately well. For aryl-containing ynol ethers, a solvent mixture of 1,2-dichloroethane (DCE) and TFE (1:1) was used.⁶ Phenyl derivative 2k gave cycloadduct 3k in 94% isolated yield and 65–80% yields were obtained for both electron-rich and electron-poor aryl derivatives. The electron-rich anisole 2n required slower addition (4 h) to overcome its hypothesized coordinative deactivation of the catalyst. Supporting this idea, slower addition of the ynol ether produced cycloadduct 3n in 80% yield (see SI, Table S1). Nitro-groups (3l), esters (3m), additional ethers (3n) and aryl-bromides (3j, 3o–3q) were also well tolerated. Bromide substitution was accommodated at all aryl positions, providing versatile handles for subsequent diversification.

Scheme 2. Substrate Scope^d.

^aAdditional 12% of double bond migration byproduct were isolated. ^bComplex product mixture was formed. ^c2n was added over 4 h. ^dReaction conditions: 5 mol % catalyst, 1.0 equiv VCP, 1.1 equiv ynol ether added dropwise over 2 h. Solvent: TFE (3a–3h), TFE/DCE 1:1 (3i–3q). For aryl substituted ynol ether (3k–3q), the reaction mixture was stirred for additional 2 h.

While many alkyl-substituted ynol ethers can be made in pure form,²⁰ their purification over silica results in substantial decomposition. The use of crude ynol ethers was therefore tested as an alternative. Two substrates (2c and 2d), purified and unpurified (see SI, Table S2), gave identical yields. The aryl substrates were more robust and were purified using triethylamine neutralized silica.

Next, catalyst loading and reaction scale were investigated (see SI, Table S3). With 5 mol % catalyst, ynol ether, 2a gave cycloadduct 3a in 87% isolated yield (Scheme 2). Significantly, a near equivalent yield (86%) was obtained with 1 mol % of catalyst. When tested on a 1 mmol scale at room temperature using 1 mol % of catalyst, 3a was obtained in 87% yield (Scheme 3). To check substrate generality, 2e was also tested, giving 3e in 93% isolated yield (Scheme 3).

Scheme 3.

Cycloaddition with Reduced Catalyst Loading at 1 mmol Scale

To explore regioselectivity, the reactivity of VCP 1b was examined. In this case, 2 equiv of VCP 1b provided improved yields. Significantly, only the 5,7-dialkyl substituted cycloadducts 4d and 4e were isolated to indicate a 1:1 mixture of diastereomers (Scheme 4).

Scheme 4.

Regioselective Access to 5,7-Disubstituted Cyclohepta-1,4-diones

Two regioisomers are possible depending on the ynol ether orientation during insertion. Previous studies have shown that alkyl-substituted terminal alkynes exhibit moderate regioselectivity (up to 7:1) using VCP 1b.²² Internal ynol ethers have not been tested previously. Providing the first experimental data on this issue of more general mechanistic and synthetic importance, ynol ethers 2d and 2e were found to react with excellent regioselectivity (>20:1).

To determine whether access to 6-substituted CHDs could also be achieved, the reaction of VCP 1c was examined. As observed with ynol ethers 2d and 2e (Scheme 4), the cycloaddition of VCP 1c and ynol ether 2a proceeded with excellent regioselectivity to give only one regioisomer, CHD 5a, in 76% yield (Scheme 5).

Scheme 5.

Regioselective Access to 6-Substituted Cyclohepta-1,4-dione

Significantly, this method is not limited to oxygen substituted VCPs. Alkyl substituted VCPs also work well, as shown by the reaction of VCP 1d with ynol ether 2a, which gave cycloheptenone 6a in 73% (Scheme 6, top). This method provides a strategically complementary route to cycloheptenones, as one can choose the more accessible VCPs and alkynes to produce a common product.^5a

Scheme 6.

Application of an Alkyl-Substituted VCP

To further test functional group tolerance, the reaction of VCP 1a with ynol ether 2k was conducted in the combined presence of acetone, ethyl acetate, diethyl ether, triethyl amine, cyclohexene and maleic anhydride (0.3 equiv of each). Using the conditions given in Scheme 2, cycloadduct 3k was isolated in 86% yield, indicating broad functional group tolerance. Prompted by these results and the previously reported preference for DCE and TFE as solvents,^21b the cycloaddition was conducted in acetone. Significantly, excellent yields were obtained in a room temperature reaction that was complete in 30 min (Scheme 7). Slow addition was not required. Acetone is thus a superb non-halogenated solvent option for both aryl- and alkyl-ynol ether substrates.

Scheme 7.

Use of Acetone As a Solvent

In summary, we report the first use of ynol ethers as ketene equivalents in the rhodium-catalyzed intermolecular [5 + 2] cycloaddition reaction with VCPs and the first study of reaction regioselectivity. The cycloaddition proceeds at room temperature within minutes to hours and provides substituted cyclohepta-1,4-diones in good to excellent yields. The reaction tolerates a wide range of functionality commonly encountered in synthesis and can be run in various solvents (DCE, TFE, acetone). Substituted VCPs can also be used and react with unprecedentedly high regioselectivity. For cost, safety and time considerations, these exploratory experiments were conducted on a small scale but are not affected by a 10-fold scale increase and can be done with a catalyst loading of 1 mol %. The use of these substituted CHDs in synthesis and as scaffolds in designed libraries will be reported in due course.