Enantioselective Construction of Spirocyclic Oxindolic Cyclopentanes by Palladium-Catalyzed Trimethylenemethane-[3+2]-Cycloaddition

The transition metal-catalyzed [3+2] trimethylenemethane (TMM) cycloaddition is a powerful and versatile method for the construction of cyclopentanes.¹ Pd-TMM complexes generated from 3-acetoxy-2-trimethylsilylmethyl-1-propene and catalytic amounts of palladium react with electron deficient olefins to produce exo-methylenecyclopentanes in a highly chemo-, regio-, and diastereoselective manner.² The ubiquity of cyclopentane containing natural products makes the development of an efficient asymmetric process highly desirable. However, applications of this methodology in asymmetric catalysis are very rare.³ Our ongoing efforts towards the synthesis of complex oxindole alkaloids prompted us to investigate the reactivity of 3-alkylidene-oxindoline-2-ones 1 towards Pd-TMM complexes.⁴ We chose the cyano-substituted TMM-precursor 2,⁵ reasoning it could enhance the asymmetric induction. Furthermore, this provides an increase in molecular complexity by the creation of an additional stereogenic center as well as installation of a synthetically valuable and versatile functionality.

The initially formed Pd-TMM complex 3a equilibrates rapidly to the most stabilized species 3b,⁶ which then adds across the activated double bond of 1 to form the exo-methylenecyclopentane (Scheme 1). Our initial experiments with hexamethylphosphorous triamide (HMPT) as ligand for palladium showed that the desired cycloadduct formed in excellent yield as a cis/trans-mixture (Table 1, entry 1). A screen to elucidate the influence of the oxindole nitrogen substituent with our previously utilized ligand L1,^3a revealed that electron withdrawing groups significantly improved the reactivity (entries 5-9). Among them, the methoxycarbonyl group provided the optimal balance between reactivity, selectivity and ease of removal. In all these cases L1 gave predominantly the cis-cyclopentane 5 (1:3 ratio for entry 9). This contrasts the reaction with HMPT as ligand (entry 1), where a 2:1 ratio favoring the corresponding trans-product 4 was observed. Our attempts to optimize the conditions for the formation of both diastereomers resulted in good diastereo- and excellent enantioselectivities for either trans-4 (95% ee and 4.3:1 dr with L3, entry 16) and cis-5 (99% ee and 1:6.2 dr with L2, entry 17).⁷ This remarkable divergence of L2 and L3,⁸ differing only by the position of their naphthyl-substituents on the pyrrolidine part of the ligand, might be rationalized as depicted in Scheme 1: The bulky 1-naphthyl-substituents of L3 preferentially orient the aromatic oxindole part of the substrate 1 under the BINOL portion of the ligand. The 2-naphthyl-substituents of L2 are shielding an area closer to the phosphorus center of the ligand, favoring an orientation of the oxindole benzene ring away from the BINOL-portion of the ligand.

Scheme 1.

Table 1.

Selected Optimization Studies.^a

entry	Ligand	T, °C	R	%Yield	4 / 5	4 %ee	5 %ee
1	HMPT	23	CO2Me	99	2:1
2	L1	23	H	-b	-	-	-
3	L1	23	Me	91	1:1	91	53
4	L1	23	PMB	60c	1:1	-	-
5	L1	23	Ts	50	1:2	-	-
6	L1	23	Boc	95	1:1.1	74	93
7	L1	23	Acetyl	98	1:2.2	85	89
8	L1	23	CO2Me	94	1:2.3	72	73
9	L1	0	CO2Me	93	1:3	80	97
11	L4	23	CO2Me	86	4.3:1	83	73
12	L4	0	CO2Me	84	4.5:1	87	60
13	L5	23	CO2Me	0	-	-	-
14	L6	23	CO2Me	99	1.1:1	27	27
15	L7	23	CO2Me	0	-	-	-
16	L3	0	CO2Me	97	4.3:1	92	95
17	L2	-20	CO2Me	97	1:6.2	96	99

^aR’ = R” = Me; all reactions were performed at 0.2 M in toluene with 2.5% Pd₂dba₃·CHCl₃, 10% ligand, 1.5 eq. 2 and stirred for 12 h; yields were combined isolated yields; ee’s were determined by HPLC with a chiral stationary phase column;

^bcomplex mixture;

^cconversion.

With these conditions elaborated, we then turned our attention towards the scope of the reaction as summarized in Table 2. Variation of the substituents of the oxindole portion had little influence on the ee (85-96% with L3 for trans-4 and 92-99% with L2 for cis-5).

Table 2.

Initial Scope of the Substituted TMM [3+2] Cycloaddition.^a

entry	substrate 1	ligand	% yield	4 / 5	4	% ee	5	% ee
1	(R = H)	L2b	97	1:6.2	4a	96	5a	>99
	1a	L3c	97	4.3:1		92		95
2	(R = 6-Cl)	L2b	90	1:2.7	4b	99	5b	92
	1b	L3c	99	19:1		93		77
3	(R = 6-MeO)	L2c	94	1:2.7	4c	95	5c	99
	1c	L3d	94	4:1		85		84
4	(R = 6,7-MeO)	L2b	96	1:3:1	4d	>99	5d	99
	1d	L3c	97	14:1		96		80
5		L2b	99	1:2.3	4e	99	5e	>99
		L3c	99	15:1		86		76
6		L2b	99	1:5:7	4f	>99	5f	>99
		L3c	91	4:6:1		94		92

^aAll reactions were performed at 0.2 M in toluene and stirred for 12 h; yields were combined isolated yields; ee’s were determined by HPLC with a chiral stationary phase column;

^bat -20°C;

^cat 0°C;

^dat 23°C.

However, the diastereoselectivity is sensitive to the benzenoid substitution pattern, wherein L2 gave optimal dr with no substitution (entry 1) and L3 gave optimal dr with the more highly substituted systems (entries 2, 4 and 5). An X-ray crystal structure analysis of 6-chloro-substituted cycloadduct 5b (entry 2) unambiguously allowed the determination of the absolute configuration to be R at C3. Exchanging the oxindole residue by the related benzofuranone 1f (entry 6) gave comparable results as illustrated by the parent substrate 1a (entry 1).

We also explored the influence of an unsymmetrical substitution pattern on both sides of the double bond, thus creating a third stereocenter in the cycloadduct (Table 3). A smooth addition was observed even with the sterically demanding trisubstituted olefins bearing a t-butyl group (1g, 1h entry 1 and 2). The observed stereochemistry at C1 and C2 of the resulting cycloadducts 4g/4h and 5g/5h reflect the double bond geometry of the substrate,⁹ supporting our proposed approach depicted in Scheme 1.¹⁰ However, the observed ee’s with L3 were only modest, compared to the reaction with L2 which still proceeded in high selectivity (entry 1 and 2). Electron rich enol ethers (1i, entry 3) are tolerated as well as substrates bearing an additional electron withdrawing ethyl ester group on the double bond (1j, entry 4).¹¹ A substrate with a fully unsymmetrical tetrasubstituted double bond formed the expected cyclopentane with two adjacent all carbon quaternary stereogenic centers (1k, entry 5).¹² With this highly hindered substrate the use of the more reactive ligand L2 was mandatory to maintain complete conversion. Interestingly - regardless of the used ligand - the trans-adduct is formed preferentially.

Table 3.

Scope of the Substituted TMM [3+2] Cycloaddition.^a

entry	substrate 1	ligand	% yield	4 / 5	4	% ee	5	% ee
1
	1g	L2b	89	1:2.6	4g	98	5g	99
		L3c	80	>20:1		20		-
2
	1h	L2d	86	3.4:1	4h	92	5h	92
		L3d	72	3.9:1		74		12
3
	1i	L2b	90	1:5.7	4j	80	5j	98
		L3c	78	3.2:1		80		41
4
	1j	L2d	72	3.1:1e	4j	87	6e	-
		L3d	60	3.4:1e		55		-
5
	1k	L2d	76	4.2:1e	4k	90	7e	-
		L3 d	34	4.0:1e		50		-

^aAll reactions were performed at 0.2 M in toluene with 2.5% Pd₂dba₃ ·CHCl₃, 10% ligand, 1.5 eq. 2 and stirred for 12 h; yields were combined isolated yields; ee’s were determined by HPLC with a chiral stationary phase column

^bat -20°C

^cat 0°C

^dat 23°C

^e4:6 and 4:7 with 6 and 7 being tentatively assigned structures.

In summary, we have demonstrated a catalytic asymmetric palladium-catalyzed [3+2] cycloaddition with cyano-substituted Pd-TMM-complexes leading to spirocyclic oxindolic cyclopentanes. Remarkably, L2 and L3 complement each other by providing the opposite diastereomers of the cycloadduct. The reaction proceeds with sterically demanding olefins under mild conditions generating arrays of up to three stereogenic centers with excellent yields and enantiomeric excesses. Further extension of the reaction scope and its application in the synthesis of complex target molecules using cycloadducts such as 4d and 4e of Table 2 are ongoing projects in our laboratory.