C2-Alkylation of N-pyrimidylindole with vinylsilane via cobalt-catalyzed C–H bond activation

Direct C2-alkylation of an indole bearing a readily removable N-pyrimidyl group with a vinylsilane was achieved by using a cobalt catalyst generated in situ from CoBr2, bathocuproine, and cyclohexylmagnesium bromide. The reaction allows coupling between a series of N-pyrimidylindoles and vinylsilanes at a mild reaction temperature of 60 °C, affording the corresponding alkylated indoles in moderate to good yields.

Over the past few years, our group and others have explored C-H bond functionalization reactions using cobalt complexes as inexpensive transition-metal catalysts [22], which often feature mild reaction conditions and unique regioselectivities [23][24][25][26][27][28][29][30][31][32]. As a part of this research program, we have recently reported a C2-alkenylation reaction of N-pyrimidylindoles with internal alkynes catalyzed by a cobalt-pyridylphosphine complex (Scheme 1a) [33], in which the pyrimidyl group functions as a readily removable directing group [34]. We also reported an ortho-alkylation reaction of aromatic imines with vinylsilanes and simple olefins using a cobalt-phenanthroline catalyst (Scheme 1b) [35]. Building on these studies, we have developed a cobalt-bathocuproine catalyst for the direct C2-alkylation reaction of N-pyrimidylindoles with vinylsilanes, which is reported herein (Scheme 1c).
Additional screening of N-heterocyclic carbene (NHC) and phosphine ligands did not lead to an improvement of the catalytic efficiency (Table 1, entries 7-9). The reaction turned out to be sensitive to the amount of the Grignard reagent, as reduction of its loading from 100 to 60 mol % improved the yield of 3aa while suppressing the formation of byproduct 4 (Table 1, entry 10).
Next, we performed screening of Grignard reagents using bathocuproine as the ligand (Table 2). Among Grignard reagents without β-hydrogen atoms, neopentyl-and phenylmagnesium bromides afforded 3aa in comparable yields ( Table 2, entries 1 and 4), while trimethylsilylmethyl-and methylmagnesium chlorides gave much poorer results ( Table 2, entries 2 and 3). Primary and secondary alkyl Grignard reagents also promoted the reaction, in which the reaction efficiency was strongly dependent on the alkyl group (Table 2, entries 5-10). We identified cyclohexylmagnesium bromide as the optimum Grignard reagent, which afforded 3aa in 69% isolated yield without formation of the cross-coupling product 4 between 1a and the Grignard reagent.
With the optimized catalytic system in hand, we explored the scope of the reaction (Scheme 2). A variety of N-pyrimidylindoles participated in the reaction with vinyltrimethylsilane to afford the alkylation products 3ba-3ia in moderate yields, with    tolerance of electron-withdrawing (F and Cl) and electrondonating (OMe) substituents and steric hindrance at the C3 and C7 positions. Unlike the cobalt-catalyzed C2-alkenylation reaction (Scheme 1a) [33], the reaction did not tolerate a cyano group on the indole substrate. In addition, N-pyrimidyl benzimidazole did not participate in the present alkylation reaction, although it was a good substrate for the C2-alkenylation reaction. A pyridyl group served as an alternative directing group to the pyrimidyl group, affording the alkylation product 3ka in 80% yield. On the other hand, an N,N-dimethylcarbamoyl group, which was previously used as a directing group for rhodium-catalyzed C2-alkenylation [36], was entirely ineffective. Vinylsilanes bearing dimethylphenylsilyl and triphenylsilyl groups were amenable to the addition reaction with 1a, affording the adduct 3ab and 3ac in modest yields. Vinyltriethoxysilane also reacted with 1a in 20% yield, although the product could not be separated in a pure form.
Unfortunately, the present catalytic system was not very effective for C2-alkylation with simple olefins. The reaction of 1a with norbornene (2d) afforded the alkylation product 3ad in 30% yield (Scheme 3a). The reaction of 1-octene (2e) was even more sluggish, affording the alkylation product 3ae in only 9% yield (Scheme 3b). Styrene also reacted rather sluggishly to afford only a small amount of the alkylation product (3% as estimated by GC and GCMS), the regiochemistry (branched versus linear) of which has yet to be determined. An acrylate ester was not tolerable as an olefinic reaction partner because of the presence of excess Grignard reagent.
The present alkylation reaction could be performed on a preparatively useful scale. Thus, alkylation of 1a with vinyltrimethylsilane (2a) on a 5 mmol scale afforded the adduct 3aa in 68% yield (Scheme 4). Furthermore, the pyrimidyl group on 3aa could be readily removed by heating with NaOEt in DMSO, affording the free indole 4aa in 85% yield.

Conclusion
In summary, we have developed a cobalt-bathocuproine catalyst for C2-alkylation of N-pyrimidyl indoles with vinylsilanes.
The reaction could be performed at a mild temperature of 60 °C, on a preparatively useful scale. Ensuing studies will focus on the development of more broadly applicable catalytic systems for the direct alkylation of indole and other heterocycles.