Rh-Catalyzed Decarbonylative Coupling with Alkynes via C–C Activation of Isatins

Abstract

Herein we report a [5+2-1] transformation though catalytic decarbonylative coupling between isatins and alkynes, which provides a unique way to synthesize 2-quinolinone derivatives. A broad range of alkynes can be coupled efficiently with high regioselectivity. This reaction is proposed to go through C–C activation of isatins, followed by decarbonylation and alkyne insertion. Directing group (DG) plays a critical role in this transformation. Assisted by the DG, the C–C cleavage of isatins occurs at room temperature.

Graphical Abstract

Cycloaddition reactions, the [X+Y(+Z)] transformation, are a powerful way to construct rings by uniting multiple π systems.¹ Alternatively, a complementary strategy is through cleavage of a C–C bond² in a cyclic compound followed by insertion of a 2π-unsaturated unit (e.g. olefins or alkynes). In particular, when cyclic ketones are employed as the substrates, C–C cleavage accompanied by CO extrusion and 2π-insertion offers a distinct approach to build ring structures, because it represents a unique [X+Y-1] transformation.^3,4 While Murakami/Ito demonstrated stoichiometric and catalytic decarbonylation of strained and unstrained cyclic ketones leading to ring contraction two decades ago,⁵ such transformations coupled with 2π-insertion have been much underdeveloped to date. The intermolecular decarbonylative couplings of cyclobutenediones and cyclobutenones with norbornene and ethylene were first reported by Kondo/Mitsudo (Scheme 1A).^6a-6b Yamamoto later developed a related intramolecular coupling between squaric acid-derivatives and olefins.^6c Very recently, our group described a decarbonylative coupling of benzocyclobutenones with alkynes to prepare fused indene compounds (Scheme 1B).⁷ While efficient, these [4+2-1] reactions are limited to highly strained four-membered ring ketones. An intriguing question is whether less strained systems, such as five or six-membered rings, can be used as the substrates for the decarbonylative C–C activation/2π-insertion reaction, which, to the best of our knowledge, has not been explored previously. Herein, as an exploratory study, we describe our development of a catalytic [5+2-1] transformation⁸ through directed C–C activation of isatins,⁹ a five-membered ring ketone, followed by decarbonylation and intermolecular alkyne insertion (Scheme 1C). This method provides a distinct strategy to synthesize various 2-quinolinone derivatives from isatins.

Scheme 1. Decarbonylative C–C activation of cyclic ketones with insertion of an unsaturated moiety.

Compared to four-membered ring compounds, one key concern of activating a less-strained C–C bond is the competing C–H activation. While a C–C σ bond is generally weaker than a C–H bond, C–H activation is often kinetically more favorable.¹⁰ As illustrated in Scheme 2, when isatin 1 is employed as the substrate, both C–H and C–C activation pathways are possible, leading to either ortho-vinylation or decarbonylative ring expansion. Given that directed addition of aryl C–H bonds across alkenes or alkynes is known to be efficient,¹¹ selective activation/subsequent functionalization of the C–C bond in isatins can be challenging.¹²

Scheme 2. Reaction pathways: C–C vs C–H activation.

(1)

To address the above challenge, we hypothesize that, through control of the orientation of the directing group (DG),¹³ it is likely that a chemoselective activation of the C–C bond can be achieved. Given that the carbonyl group is significantly larger than the hydrogen, increasing steric hindrance on the “back” site of the DG should minimize the conformation that leads to C–H activation (e.g. Eq 1). To examine this hypothesis, two DGs with different steric properties were evaluated initially (Eq 2). While both provided the desired 2-quinolinone products, the 3-methyl-2-pyridyl group¹⁴ (isatin 1a) showed superior selectivity compared to the corresponding 2-pyridyl group (1b). For example, under the optimized reaction conditions (“standard conditions”), isatin 1a afforded the desired decarbonylative ‘cut and sew’ product 3aa in 88% yield without any observable C–H activation products; in contrast, with 2-pyridyl as the DG, a significant amount of C–H vinylation (including 4% C–H-activation only and 3% sequential C–C/C–H-activation products)^1,5 and decomposition products were formed.

Control experiments were subsequently conducted to understand the role of each reactant (Table 1). In the absence of the Rh catalyst, no desired product was observed (Entry 2). Use of Wilkinson's catalyst [Rh(PPh₃)₃Cl] instead gave only 8% of the desired product together with high recovery of the starting material (Entry 3). Other Rh(I) catalysts, such as [Rh(coe)₂Cl]₂, [Rh(C₂H₄)₂Cl]₂ and [Rh(CO)₂Cl]₂, showed lower efficiency (Entries 4-6). Employment of other metals, such as Ru₃(CO)₁₂ and Co₂(CO)₈, didn't afford any desired product (Entries 7 and 8). Addition of phosphine ligands all inhibited the reaction to various extents (Entries 9-11). Solvent effect was also surveyed: toluene proved less effective (Entry 12); DCE completely decomposed the starting material (Entry 13); dioxane worked almost equally well as chlorobenzene affording 2-quinolinone 3aa in 84% yield (Entry 14). Decreasing the temperature to 150 °C resulted in lower conversion (Entry 15).

Table 1. Control experiments^a.

entry	change from standard conditionsc for 1a	conversion	yield of 3aa
1	None	>99%	88%
2	Without Rh catalyst	14%	-
3	Rh(PPh3)3CI instead of [Rh(cod)CI]2	22%	8%
4	[Rh(coe)2CI]2 instead of [Rh(cod)CI]2	93%	73%
5	[Rh(C2H4)2CI]2 instead of [Rh(cod)CI]2	73%	69%
6	[Rh(CO)2CI]2 instead of [Rh(cod)CI]2	73%	62%
7	Ru3(CO)12 instead of [Rh(cod)CI]2	73%	trace
8	Co2(CO)8 instead of [Rh(cod)CI]2	14%	-
9	Additional 20 mol% of PCy3 was added	75%	23%
10	Additional 20 mol% of P(C6H3(CF3)2)3 was added	24%	-
11	Additional 20 mol% of dppb was added	40%	15%
12	Toluene instead of PhCI	57%	29%
13	DCE instead of PhCI	>99%	-
14	1,4-Dioxane instead of PhCI	>99%	84%
15	Temperature was 150 °C	56%	43%

^aAll yields and conversions were determined by crude ¹H NMR using 1,1,2,2-tetrachloroethane as the internal standard.

^b4% of ortho-C–H vinylation product and 3% of sequential C–H/C–C activation product were observed.

^cThe standard reactions were conducted with 1a (0.04 mmol), 2a (0.048 mmol, 1.2 equiv), [Rh(cod)Cl]₂ (0.002 mmol, 5 mol%), and PhCl (1 mL) in a 4 mL vial at 160 °C. ^c

The reaction scope was first tested using substrate 1a with different alkynes under the optimized conditions (Table 2). Both aryl and alkyl substituted internal alkynes with various electronic properties all worked very well affording high yields of the corresponding 2-quinolinones. Heterocycles, such as pyridine (2d) and thiophene (2e), are tolerated; unprotected alcohols, esters and ketones are also compatible due to the pH and redox neutral conditions. For unsymmetrical internal alkynes, good to excellent regioselectivity was observed. It is interesting to note that the major regioisomer tends to have the more sterically hindered substituent at the 4-position of the 2-quinolinone, though the exact reason is unclear. The structures of the major regioisomers were further confirmed by 1D-NOESY study and/or X-ray crystallography. Terminal alkynes,¹⁶ such as 1-decyne 2h, can also be used as the coupling partner albeit giving a relatively lower yield (50% for the major regioisomer) and a moderate selectivity (6/1).

Table 2. Substrate scope with different alkynes^a.

^aAll yields are isolated yields; regioselectivity of the products was determined by crude ¹H NMR.

^bIsolated yield for the major regioisomer only.

The scope of the isatins and DGs was next investigated and compared (Table 3). Besides pyridine-based groups, other DGs are also effective, such as 1H-indazol-1-yl (3ca) and oxazolinyl groups (3da and 3ha).¹⁷ As expected, while no ortho-C–H activation was observed with the 1H-indazol-1-yl DG, about 6% yield of ortho-C–H vinylation product was found with the less hindered oxazolinyl DG. Isatin substrates with a methyl group at C-5 position led to the corresponding product 3ea in 92% yield. In addition, the N-benzyl protected isatin worked equally well (3fa). Finally, the 7-aza-isatin compound was also found to be a viable substrate, giving an interesting 1,8-naphthyridinone (3ga) as the product.

Table 3. Substrate scope with different amines^a.

^aAll yields are isolated yields.

^b4% of ortho-C–H vinylation product and 3% of sequential C–H/C–C activation product were observed.

^c6% of ortho-C–H vinylation product was observed.

^d2.0 equiv of alkyne 2a was used.

^eThe reaction conversion was 83%, determined by crude ¹H NMR.

Regarding the mechanistic features, a key question is why the reaction requires a relatively high temperature to take place? Consequently, it is natural to be concerned about the ease of oxidative addition of Rh into the isatin C–C bonds. To probe the mechanism, reaction of isatin 1a with stoichiometric [Rh(C₂H₄)₂Cl]₂ was investigated (Scheme 3). Interestingly, the C–C cleavage occurred smoothly even at room temperature (rt).¹⁸ The red color of isatin disappeared and a light yellow solid slowly precipitated, which is presumably a chlorine-bridged oligomer (or dimer). This Rh complex was then treated with excess pyridine, affording a mixture of two diastereomers (M1 and M2), which are more soluble and can be separated using silica-gel flash chromatography. Another reason of adding pyridine is to mimic the real reaction conditions in which excess substrate (pyridine-like) is present. Both M1 and M2 are expected to be the products after C–C activation and CO deinsertion, the structures of which were confirmed through NMR, IR and HRMS. The Rh-CO stretching frequencies for M1 and M2 are 2067 and 2044 cm^-1 respectively. In addition, the structure of M1 was further unambiguously confirmed by X-ray crystallography.¹⁹ It is worthy to note that when the parent N-methylisatin (without DG) was used as the substrate, no C–C activation product was observed under either the catalytic or stoichiometric-metal conditions.²⁰

Scheme 3. Mechanistic exploration.

Reaction of complex M1 and M2 with alkyne 2a was next studied. It is not surprising to find that alkyne insertion did not occur when the temperature was below 130 °C. Nevertheless, at 160 °C the quinolinone product (3aa) was isolated in 59% yield. Similar experiments have also been conducted using the C–C cleavage complex (the precursor for M1/M2) in the absence of added pyridine. Again, its decarbonylative coupling with alkyne 2a showed no significant reactivity below 130 °C. In addition, complex M1 was found to catalyze the reaction giving a good yield of the product (Eq 3).

(3)

From these studies, the following information can be obtained. First, the observations are consistent with the proposed mechanism involving directed C–C activation (Scheme 2, left cycle). Second, the DG plays a critical role in assisting the C–C activation. Third, there is a remarkable reaction-temperature difference between the C–C activation and the following steps in stoichiometric metal settings. More in-depth mechanistic study is still ongoing.

In summary, we have developed a Rh-catalyzed [5+2-1] transformation through catalytic C–C activation of isatins followed by decarbonylative coupling with alkynes. A range of isatin and alkyne substrates are suitable for this reaction providing the corresponding poly-substituted 2-quinolinones in good to excellent yields. The reaction conditions are both pH and redox neutral, which is the key for the tolerance of various functional groups and heterocycles. The activation of the C–C bond in a relatively less strained cyclic ketone followed by 2π-insertion should have broad implications beyond this work. Efforts on expanding the reaction scope with alkenes and developing C–C activation of less strained ketones in the absence of auxiliary DGs are currently undertaken in our laboratory.