Rh^I-Catalyzed Decarbonylative Spirocyclization via C–C Cleavage of Benzocyclobutenones: An Efficient Approach to Access Functionalized Spirocycles

Abstract

We describe a rhodium-catalyzed all-carbon spirocenter formation through a decarbonylative coupling of trisubstituted cyclic olefins and benzocyclobutenones via C–C activation. A [Rh(CO)₂Cl]₂/P(C₆F₅)₃ metal-ligand combination was found to catalyze this transformation most efficiently. A range of diverse spirocyclic rings were synthesized in good to excellent yields and many sensitive functional groups were tolerated. Mechanistic study supports the hydrogen-transfer process that occurs via a β-H elimination/decarbonylation pathway.

Catalytic carbon–carbon bond (C–C) activation/functionalization provides a unique platform to develop novel transformations that are otherwise challenging using conventional approaches.^[1] In particular, C–C activation that involves a decarbonylation process is of significant synthetic value because it allows for the preparation of compounds that lack a carbonyl moiety from a more readily available ketone precursor.^[2] Given the ubiquity of carbonyl compounds, the decarbonylative C–C activation followed by new C–C formation process would potentially offer a distinct synthetic strategy using carbonyl groups as a “traceless handle”.

(1)

(2)

Although highly attractive, the decarbonylative C–C activation strategy has been largely limited to its reaction scope. Previous efforts primarily involved a direct CO extrusion to give the corresponding hydrocarbon products (eq 1).^[3,4] Only a few examples are known to engage the addition of simple olefins (eq 2).^[5] To advance the scope and applicability of the decarbonylative C–C activation/functionalization strategy, new classes of synthetically useful transformations are highly sought. Herein we describe our development of a rhodium-catalyzed decarbonylative spirocyclization reaction through an intramolecular coupling between olefins and benzocyclobutenones (Figure 1), in which the reaction course can be controlled by the choice of the ligand on the metal catalyst.

Figure 1.

Rh^I-catalyzed decarbonylative coupling between olefins and benzocyclobutenones via C–C activation.

Recently, we developed a Rh-catalyzed intramolecular carboacylation between benzocyclobutenones and olefins.^[6] This “cut and sew” transformation begins with the oxidative addition of Rh^I into the benzocyclobutenone C₁–C₂ bond, which is followed by migratory insertion into the olefin to give a seven-membered metallocycle (C), and subsequent reductive elimination to afford fused-ring systems (D). However, if a β-H elimination/decarbonylation process can be promoted instead of direct C–C reductive elimination, spirocycles containing all-carbon quaternary centers would be rapidly constructed when cyclic trisubstituted olefins were employed as the coupling partner (Figure 1). Given that spirocycles are important structural motifs often found in bioactive natural products whereas efficient synthesis of functionalized spirocycles has heretofore been challenging,^[7] this decarbonylative C–C activation strategy would provide a complementary approach to the previous spirocyclization methods.^[8] However, the challenge here is how can one promote the β-H elimination and decarbonylation instead of direct reductive elimination?

(3)

Previously, we discovered that bidentate phosphine ligands bearing a large bite-angle promoted direct reductive elimination; however, a Lewis acid is needed to enhance the electrophilicity of the substrate to permit coupling with poly-substituted olefins (eq 3).^[6a] We hypothesized that the use of monodentate π-acidic ligands (inspired by a recent work of Tang^[9]) would benefit spirocycle formation in two different aspects: 1) the faster ligand exchange (compared to bidentate ligands) would facilitate forming open coordination sites on the Rh, which in turn could favor both β-H elimination and CO deinsertion; 2) the more electron-deficient catalyst would coordinate more strongly with the trisubstituted olefins thus enhancing subsequent migratory insertion.^[10]

To test our hypothesis, compound 1a that previously gave the tetracyclic “cut and sew” product (2a, eq 3) was employed as the model substrate for the spirocycle formation. A number of Rh^I precatalysts and phosphine ligands were examined. Indeed, when electron-rich or bidentate ligands, such as PCy₃, dppb and dppf, were employed, no desired decarbonylative spirocyclization product was obtained; in contrast, use of the more electron-poor [Rh(CO)₂Cl]₂ alone produced the 2H-benzofuran-[4.5]-spirocycle (3a) in ca 3% yield (entry 1, Table 1). Use of an acac ligand on the Rh versus Cl was found to be detrimental to the catalyst reactivity, leading to slight decomposition of 1a (entry 2, Table 1). The in situ generated cationic Rh^I led to the dealkylation of 1a to give 3-OH-benzocyclobutenone (entry 3, Table 1). Use of the more electron-rich PPh₃ or highly electron-deficient phosphites as the ligands completely shut down the catalyst reactivity (entries 4–6, Table 1). However, we found that employment of the π-acidic triarylphosphine ligands significantly promoted formation of the desired spirocycle 3a (entries 7–11, Table 1), among which the P(C₆F₅)₃ ligand proved to be most efficient. Finally, simply by lowering the ligand/metal ratio to 1:1, formation of the undesired reductive-elimination product (2a) was significantly inhibited; spirocycle 3a was isolated as the major product in a 72% yield (entry 10, Table 1). Presumably, when less ligand is present, the metal tends to provide open coordination sites for β-H elimination.^[10] It is interesting to note that under these reaction conditions the C₃-olefin isomer (after one chain-walk^[11]) of the spirocycle product was selectively afforded (for a mechanistic study, vide infra).^[12,13] Further lowering the ligand/metal ratio to 0.5:1 gave a similar result (entry 11, Table 1).

Table 1.

Selected optimization conditions

Entry	Precatalyst	Ligand/Additive	Yield[a]
			3a/4a	2a
1	[Rh(CO)2Cl]2	none	3%/<1%	2%
2[b]	Rh(CO)2acac	none	0%	0%
3[c]	[Rh(CO)2Cl]2	10 mol % AgSbF6	0%	0%
4	[Rh(CO)2Cl]2	PPh3	0%	0%
5	[Rh(CO)2Cl]2	P(OCH2CF3)3	0%	0%
6	[Rh(CO)2Cl]2	P[OCH(CF3)2]3	0%	0%
7	[Rh(CO)2Cl]2	P(2-furyl)3	7%/≤4%	22%
8	[Rh(CO)2Cl]2	P[3,5-(CF3)2C6H3]3	14%/14%	39%
9[d]	[Rh(CO)2Cl]2	P(C6F5)3	30%/7%	60%
10 [[d],[e]]	[Rh(CO)2Cl]2	P(C6F5)3	72% [f] /≤4%	22% [f]
11[[d],[g]]	[Rh(CO)2Cl]2	P(C6F5)3	66%/≤5%	15%

^[a]Yields were determined by ¹H-NMR using mesitylene as the internal standard.

^[b]10 mol % Rh(CO)₂acac was used.

^[c]3-OH-benzocyclobutenone was isolated in 53% yield.

^[d]The reaction time was 36 h.

^[e]10 mol % P(C₆F₅)₃ ligand was used.

^[f]Isolated yield.

^[g]5 mol % P(C₆F₅)₃ was used. Acac=acetylacetonates.

The scope of this decarbonylative spirocyclization was investigated (Table 2). First, cyclic olefins with different ring sizes were examined. To our delight, 5, 6, 7, 8 and 12-membered ring substrates all underwent this transformation smoothly (entries 1–5, Table 2); a single olefin isomer was observed except for the 5-membered ring substrate (entry 2, Table 2). Intriguingly, spirocyclization of the 12-membered ring substrate (1e) proceeded to give a trans-olefin in 90% yield without further alkene isomerization (entry 5, Table 2). The enhanced reactivity of substrate 1e is likely attributed to a transannular interaction caused by the 12-membered cycle.^[14] The structures of spirocycles 3a, 3c and 3d were unambiguously confirmed by X-ray crystallography. Electron-deficient olefins, such as an enone, also reacted, albeit with a lower conversion (entry 6, Table 2). It is noteworthy that this spirocycle formation method is highly chemoselective and a variety of sensitive functional groups, including dienes, ketones, enamides, esters, benzyl and vinyl ethers, and unprotected tertiary alcohols are tolerated (entries 6–10, 15, 16, Table 2), which is likely attributed to the near neutral reaction conditions. For example, when dihydrobenzopyran 1h was employed as the substrate, the vinyl ether-based spirocycle was isolated in 71% yield (entry 8, Table 2).

Table 2.

Substrate scope^[a]

^[a]Each reaction was run on a 0.2 mmol scale in a sealed vial, using 5 mol % [Rh(CO)₂Cl]₂ and 10 mol % P(C₆F₅)₃, in THF, 130 °C, 36 h.

^[b]Isolated yields.

^[c]Ratio of the olefin regioisomers were determined by ¹H-NMR or based on their isolated yields (see supporting information for more details).

^[d]Numbers in the parenthesis are brsm yields.

^[e]Compound 3k was characterized via a subsequent olefin hydrogenation (see supporting information for more details). brsm=based on recovered starting material.

^[f]Isolated as inseparable isomers.

In addition, when the C₈-methoxy substituted benzocyclobutenone 1j was utilized, the spirocycle containing a benzyl ether moiety (3j) was isolated in 59% yield as a single isomer (Table 2, entry 10). While the exact reason is still unclear, the presence of the C₈-methoxy group likely promotes a faster C–H reductive elimination rather than olefin migration.^[10] Interestingly, the demethoxy spirocycle (3b) was also isolated in 18% yield.^[15] Furthermore, we discovered that, besides forming benzofuran-based products, other classes of spirocycles can also be efficiently synthesized: first, this transformation works well with substrates lacking an ether linkage (entry 11, Table 2); second, the 6-membered ring forming cyclization proceeded equally well to give benzopyrane-based spirocycle 3l in 79% yield (entry 12, Table 2). Note that, except for substrates 1a, 1i and 1m,^[16] no direct reductive elimination (the “cut and sew”) product was observed for other substrates depicted in Table 2. In addition, substrates with different substitutions at the C₄ and C₅ positions of the benzocyclobutenone also proceeded the spirocyclization smoothly (entries 13–16, Table 2).^[17] It is encouraging to note that substitutes with different steric and electronic properties, including methyl, 1-butyl-1-hydroxylpentyl and methyl ester groups, all provided good yields of the desired spirocycles.^[17b]

(4)

To provide mechanistic insights for this transformation, a deuterium-labeling experiment (eq 4) was designed. Substrate 1n was readily synthesized from propargyl alcohol and D₂-paraformaldehyde in five steps with an overall yield of 44% (for details, see supporting information). Subjecting 1n to the standard decarbonylative spirocyclization conditions, spirocycle 3n with >95% deuterium-incorporation at the methyl group was isolated in 79% yield. This experiment strongly supports our hypothesis that this transformation underwent a β-H elimination, decarbonylation and then C–H reductive elimination pathway (vide supra, Figure 1).

Another mechanistic puzzle regards the selective olefin chain-walk with most substrates. It was postulated that two reaction pathways are possible for the olefin migration: 1) alkene isomerization after the decarbonylative spirocyclization [i.e. the C₃-isomer comes from the C₂-isomer]; 2) alkene isomerization during the decarbonylative spirocyclization [i.e. the C₃- and C₂-isomers are formed independently]. Compounds 3i and 3i' were chosen as the model compounds because they can be readily separated through column chromatography. When pure 3i and 3i' were subjected to the standard reaction condition (shown in Table 2) respectively, no isomerization for either product was observed after 36 h (eq 5), which indicates the [Rh(CO)₂Cl]₂/P(C₆F₅)₃ system cannot isomerize the alkene in the products.

(5)

(6)

(7)

(8)

A cross-over experiment was also carried out to examine whether the intermediates generated in the catalytic cycle, e.g. a Rh^III-H species, can isomerize the alkenes. Subjection of compounds 1b and 3i under the standard reaction conditions provided 3b in 80% yield but no isomerization of 3i was found (eq 6). When the reactions of 1i and 1b were monitored separately over time, the ratio between the C₂ and C₃ isomers remained largely unchanged (eqs 7 and 8), which is consistent with our earlier observation for substrate 1a.^[13] Combining with the deuterium-labeling experiment (eq 4), these studies suggest that the olefin chain-walk likely occurs during (instead of after) the process of the spirocycle formation, thus the C₂ and C₃ isomers are expected to form independently (not shown in Figure 1).

(9)

Furthermore, we found this transformation is not limited to cyclic trisubstituted olefins, as the linear disubstituted olefins also underwent the decarbonylative cyclization to provide the cyclization products. While coupling of linear tri-substituted alkenes proved challenging,^[18] reactions with the 1,2-disubstituted olefins proceeded smoothly (eq 9).^[19] It is interesting to note that the olefin geometry does not significantly affect the reactivity.

In conclusion, we have developed a Rh-catalyzed decarbonylative spirocyclization of olefins and benzocyclobutenones via C–C activation, which provides a complementary but distinct way to generate all-carbon spirocenters. This reaction exhibits several key features. First, it operates at near neutral reaction conditions, and therefore tolerates many acid or base-sensitive functional groups, which represents a major advantage over classical spirocycle syntheses. Second, it has a broad substrate scope (both electron-rich and -poor olefins react) providing a range of structurally diversified spirocycles, which indicates great potential for application in complex molecule synthesis. Furthermore, to the best of our knowledge, this represents the only report that couples C–C activation, olefin insertion, β-H elimination and decarbonylation into a one-reaction sequence. The unique mode of reactivity described here may provide broad implications for designing new tandem transformations. Further expansion of the substrate/reaction scope and the development of an enantioselective version of this reaction^[20] are ongoing.