Palladium-Catalyzed β-Allylation of 2,3-Disubstituted Indoles

Abstract

Given the prevalence of the indole nucleus in biologically active compounds, the direct C3-functionalization of 2,3-disubstituted indoles represents an important problem. Described is a general, high-yielding method for the palladium-catalyzed β-allylation of carba- and heterocycle fused indoles, including complex natural product substrates.

Indoles and indole-derived heterocycles are prevalent structural motifs in natural products, medicinal compounds, and organic materials.¹ Given their importance, much effort has been directed toward the development of methods for the selective functionalization of the indole nucleus at the N1, C2, and C3 sites.² A particularly difficult transformation is the electrophilic attack at C3 on 3-substituted indoles to produce indolenines containing a new carbon-carbon bond and a quaternary stereocenter. Indolenine units can be found within natural products, as biogenetic precursors to indole alkaloids, and as pivotal intermediates in total syntheses.^3,4 Traditional methods for β-functionalization of indoles, which take advantage of the ambident character of indoles or indole anions, have limitations.⁵ Not only is alkylation of C3-substituted indoles difficult, but the strong, nucleophilic bases (e.g. Grignard reagents, NaNH₂) typically used for such transformations restrict their scope due to functional group incompatibility. In the course of our studies toward the total synthesis of complex alkaloids, we required a mild, functional group-tolerant method for the introduction of an allyl group regioselectively on a β-carboline substrate (e.g. 1, Scheme 1). While the literature records methods for the palladium-catalyzed β-allylation of simple indoles,⁶ the direct allylation of hindered, 2,3-disubstituted indoles, such as 1, remains an unsolved problem.⁷ We report here a general, high-yielding method for the palladium-catalyzed allylation of carba- and heterocycle fused indoles, including highly functionalized complex natural product substrates.⁸

Scheme 1.

In initial studies, we examined the effectiveness of published methods for the palladium-catalyzed indole allylation using 1,2,3,4-tetrahydrocarbazole (3) as a challenging substrate and allyl methyl carbonate (4) as the allyl source (Table 1, Entries 1-4).^6,7,9 The modest yields obtained using these protocols, prompted us to explore other catalyst systems for this transformation.¹⁰ Some of the many conditions examined during this optimization process are shown in Table 1. Pd₂(dba)₃ was quickly determined to be a suitable source of the catalytically active palladium species (Entries 5-11). With regard to phosphine ligands, hindered alkyl phosphines gave the allylated product in fair to good yields, whereas triphenylphosphine (Entry 5) and trifurylphosphine (Entry 11) gave the best yields of 5. Significantly, the rate of the reaction was found to considerably faster with trifurylphosphine than with triphenylphosphine.¹¹ Additionally, under these conditions, none of the N-allylated product was observed. Reduction in the amount of P(2-furyl)₃ used to a 1:1 ratio with palladium did not appreciably affect the yield or rate of the reaction (Entry 12). Given the faster rate, lower catalyst loadings were examined. When the reaction was carried out using 1 mol % of Pd₂(dba)₃ (Entry 13), the product was obtained in high yield, but required longer for the reaction to go to completion. Further reduction of Pd₂(dba)₃ to 0.5 mol % proved less satisfactory, giving 5 in 66% yield, even after 2 days (Entry 14). Allyl acetate (Entry 15) was found to be an ineffective allylation precursor, giving the desired product in a reduced yield and accompanied with the N-allylation product (19%). This survey defined the optimum conditions for the β-allylation of hindered indoles as shown in entry 12, which are convenient and mild.

Table 1.

Optimization of allylation reaction

	[Pd] (mol %)	PR3 (mol %)	time	yield
1a	Pd(acac)2 (2.0)	PPh3 (2.0)	20 h	17%
2b	Pd(PPh3)4 (5.0)	-	20 h	20%
3c	Pd2(dba)3 (2.5)	Trost’s ligand (7.5)	20 h	52%
4d	[PdCl(π-allyl)]2 (5.0)	dppe (11)	20 h	90%
5	Pd2(dba)3 (2.5)	PPh3 (15)	20 h	91%
6	Pd2(dba)3 (2.5)	tBu3P (15)	20 h	trace
7e	Pd2(dba)3 (2.5)	P(tBu)2(biphenyl) (15)	20 h	85%
8e	Pd2(dba)3 (2.0)	rac-BINAP (6.0)	20 h	45%
9	Pd2(dba)3 (2.5)	dppp (2.5)	20 h	NR
10	Pd2(dba)3 (2.5)	P(OMe)3 (5.0)	20 h	trace
11	Pd2(dba)3 (2.5)	P(2-furyl)3 (15)	2 h	99%
12	Pd2(dba)3 (2.5)	P(2-furyl)3 (5.0)	2 h	99%
13	Pd2(dba)3 (1.0)	P(2-furyl)3 (2.0)	20 h	92%
14	Pd2(dba)3 (0.05)	P(2-furyl)3 (1.0)	48 h	66%
15f	Pd2(dba)3 (2.5)	P(2-furyl)3 (15)	20 h	20%

^aRun in AcOH at 70 °C with allyl acetate in place of 4.

^bAllyl alcohol and Et₃B were used in place of 4.

^cAllyl alcohol and n-hexyl-9-BBN were used in place of 4; Trost’s phosphine ligand is described in ref 7b.

^dRun at 40 °C in the presence of Li₂CO₃.

^eRun in PhMe.

^fAllyl acetate was used in place of 4.

The reaction conditions developed above were found to be broadly applicable to a wide variety of carba- and heterocycle fused indoles (Table 2). Substitution the 6-position of tetrahydrocarbazole is well tolerated (Entries 1-3), although chloride 6c required 20 h for complete conversion. Cycloheptane and cyclooctane fused indoles 6d and 6e participate in the reaction and give high yields of the corresponding allyl indolenine products (Entries 4-5). Tetrahydro-γ-carbolines are competent substrates as well, but generally require longer reaction times than their all carbon counterparts. Electron rich γ-carbolines (6f-6h) were allylated in greater than 90% yield, whereas the more electron-deficient chlorocarboline 6i reacted more sluggishly, generating the allylation product 7i in 76% yield after 48 h. Tetrahydro-β-carbolines were evaluated also. Boc-protected β-carbolines 6j and 6k provided the allylated products in high yield and exhibited similar rates of reaction as the γ-carbolines. Even the more electron-deficient dihydro-β-carboline 6l was allylated at room temperature, albeit more slowly and in lower yield 63% yield. Allylation of simpler, monosubstituted indoles gave the expected allyl-indolenines in good yields (Entries 13, 14). Finally, substituted tetrahydro-β-carboline 6o gave the corresponding allylation product in 73% yield, as a 1.2:1 ratio of diastereomers.

Table 2.

Scope of Pd-catalyzed β-allylation of indoles^a

	substrate	product	time	yield
1		7a	2 h	98%
2		7b	4 h	88%
3		7c	20 h	87%
4		7d	4 h	89%
5		7e	4 h	91%
6		7f	20 h	91%
7		7g	20 h	94%
8		7h	20 h	91%
9		7i	48 h	76%
10		7j	20 h	94%
11		7k	20 h	89%
12		7l	120 h	63%
13		7m	48 h	88%
14		7n	20 h	84%
15		7o	20 h	73%

^aReactions were carried out on 0.250 mmol indole substrate with 0.025 equiv Pd₂(dba)₃, 0.050 equiv P(2-furyl)₃, and 2.0 equiv allyl methyl carbonate (4) in 2.5 mL CH₂Cl₂ at rt.

^b0.300 equiv P(2-furyl)₃ were used.

The high efficiency of the allylation reactions described above prompted us to extend the reaction to substituted allyl carbonates. The results of these studies are summarized in Table 3. Methallyl carbonate 8a reacted with carbazole 3 to give 9a in 98% yield (Entry 1). The corresponding reaction with crotyl carbonate 8b proceeded cleanly to afford indolenine 9b in high yield (Entry 2). Notably, the regio- and geometric alkene isomer products were observed in only trace amounts. The prenylation of 3 with prenyl carbonate 8c was much slower than analogous allylation reactions, but did provide the prenylated indolenine 9c in 45% yield (Entry 3). Lastly, the reaction of carbazole 3 with cinnamyl carbonate 8d afforded 9d in 92% yield as the sole observed isomer (Entry 4).

Table 3.

Allylation with substituted allyl carbonates^a

	carbonate	product	time	yield
1			20 h	98%
2			20 h	90%
3b			48 h	45%
4			20 h	92%

^aReactions were carried out on 0.250 mmol 3 with 2.5 mol % Pd₂(dba)₃, 5.0 mol % P(2-furyl)₃, and 2.0 equiv 8 in 2.5 mL CH₂Cl₂ at rt.

^bReaction was carried out in dichloroethane heated to reflux with 5.0 mol % Pd₂(dba)₃ and 10 mol % P(2-furyl)₃.

The capability of a method is assessed best when it is utilized for complex molecules containing multiple functional groups. With that consideration in mind, we examined the allylation of three different indole alkaloids of differing functional group complexity (Scheme 2). Subjection of synthetic, unprotected (±)-geissoschizol (10)¹² to the allylation protocol afforded the corresponding allyl-indolenine (11) in high diastereoselectivity in 45% yield (Eq. 1).^13,14 The allylation of yohimbine (12) was also highly diastereoselective, providing a single allylation product (13) in 79% yield (Eq. 2).¹⁵ An even more intricate molecule, reserpine (14), underwent allylation under the standard conditions and provided roughly equal amounts of the two diastereomeric allylated products 15 (Eq. 3). The relative stereochemistries of the allylation products have been tentatively assigned based on their NMR spectral properties. The dramatic differences in levels of diastereoselection for the three alkaloids are not immediately apparent. The allyl derivatives represent novel analogs of these bioactive indole alkaloids.

Scheme 2.

Allylation of complex natural products

In summary, we have described a convenient, high-yielding method for the regioselective allylation of carbazole and carboline substrates, which can be extended to substituted allyl analogs as well as to complex indole alkaloids.¹⁶ Studies directed toward the implementation of an asymmetric variant of the above reaction are underway and the results will be detailed in due course.