Synthesis of 3,3-Disubstituted Heterocycles by Pd-Catalyzed Arylallylation of Unactivated Alkenes

Finding new methods of carbon–carbon bond formation is a key goal in expanding current methodology for heterocycle formation. Because of their inherently nonplanar shape, new methods of forming sp3-rich scaffolds are of particular importance. Although there are methods for combining heterocyclization and formation of new sp3–sp3 carbon–carbon bonds, these form the carbon–heteroatom bond rather than a carbon–carbon bond of the heterocycle. Here, we show a new alkene arylallylation reaction that generates a heterocycle with concomitant formation of two new carbon–carbon bonds. Furthermore, we demonstrate that this process occurs through an isohypsic (redox neutral) mechanism. Overall, this carboallylation reaction gives a new route to the synthesis of 3,3-disubstituted heterocycles.


INTRODUCTION
Heterocycles are found in a vast array of biologically active molecules and complex natural products. Consequently, many synthetic strategies have evolved to access these motifs. 1 Palladium-catalyzed methods are among the most widely used, due to the versatility and functional group tolerance of these cycles. 2 Most Pd-catalyzed heterocycle forming reactions take advantage of the nucleophilicity of the heteroatom as a key feature of the overall process. For example, we have previously shown that difunctionalisation of alkenes by heteroallylation can be used to generate a range of O-and N-containing heterocycles ( Figure 1a). 3 A lesser-used set of reactions utilizes starting materials where the carbon−heteroatom bonds are already in place. 4 For example, Pd-promoted cyclization of aryl bromide 3, followed by Sonogashira−type coupling affords oxindole 4 (Figure 1b). 5 The inclusion of more sp 3 -hybridized carbons in medicinal chemistry programmes has been shown to correlate to increased clinical success. 6 As a complement to our earlier alkene heteroallylation work that generates heterocycles substituted with a new C(sp 3 )−C(sp 3 ) bond at the 2 position (Figure 1a), we became interested in developing a Pdcatalyzed synthesis of 3,3-disubstituted heterocycles with accompanying C(sp 3 )−C(sp 3 ) bond formation (Figure 1c). The success of this transformation would provide a new potential strategy to access complex biologically active compounds, such as the nM norepinephrine reuptake inhibitor daledalin, or the neurokinin receptor antagonist 7 (Figure 2). 7 A single literature report describes Pd-catalyzed heterocycle synthesis from aryl boronic acids (8 → 9, Scheme 1). 8 This process likely proceeds through a Pd(0)−Pd(II) catalytic cycle, including β-hydride elimination and reoxidation of the Pd by oxygen. On the basis of this precedent as well as our experience on developing an isohypsic (redox neutral) heteroallylation of alkenes (Figure 1a), we set out to develop an alkene carboallylation reaction that would generate 3,3-disubstituted heterocycles with concomitant sp 3 −sp 3 (Figure 1c).

RESULTS AND DISCUSSION
2.1. Optimization. An initial attempt at carboallylation was carried out using conditions similar to the heteroallylation (Table 1). Pleasingly, this resulted in formation of the desired dihydrobenzofuran 11, albeit as a minor component, with the main products being that of direct allylation 12 and deboronation 13 (entry 1). A change of the allyl halide from bromide to chloride resulted in decreased formation of the deboronation product 13 (entry 2). Lowering the number of the equivalents of allyl chloride caused an improvement in the ratio of 11:12, although the formation of 13 was also increased (entries 3 and 4). Continuing with 2 equivalents of allyl chloride, the effect of the catalyst system was examined ( Table  2). The use of phosphine ligands proved detrimental, with none of the desired product formed (entries 2 and 3). Palladium(II) catalyst PdCl 2 led to the slightly decreased formation of the desired product, along with lower formation of deboronation product 13 (entry 4), whereas palladium(0) catalysts did not result in formation of the desired product (entries 5−7). The use of phosphonite catalyst 14 9 resulted in an improved ratio between the desired and direct coupling products, but use of this catalyst was discontinued because of the formation of several additional unidentified side products (entry 8).
Moving from an aqueous base to a solid base greatly improved the ratio between 11:12 (Table 3, entries 1 and 2). Use of other solid bases, such as sodium hydroxide or cesium fluoride, caused lower or zero conversion to the desired product (entries 3 and 4). Use of organic bases, such as triethylamine, also resulted in formation of the desired product as a minor component (entry 5).
Next, the choice of solvent was examined ( Table 4). Use of dimethylformamide (DMF) showed no conversion to the desired product, with the only product being that of direct allylation (entry 2). Acetonitrile and tetrahydrofuran (THF) not only lowered the formation of the deboronated product but also resulted in the formation of less desired product (entries 3 and 4). Use of dimethoxyethane completely suppressed the formation of the deboronated product, although overall conversion was lowered (entry 5).
Carrying these conditions forward, the choice of boron reagent was examined (Table 5). Pinacol boronate ester proved ineffective, with only a small amount of the desired product formed, with direct allylation product 12 being the major result (entry 2). N-methyliminodiacetic acid (MIDA) boronate showed very low conversion, with the only product formed that of direct allylation (entry 3). Potassium   As the competition between desired 11 and direct coupling 12 products involves a cyclization step, the "reactive rotomer" effect 10 can have an important influence on product ratio (Scheme 2). After transmetallation to form Pd(II) intermediate 16, the proximity of the O-methallyl chain determines how easily direct coupling or cyclization can occur. For cyclization to occur, the O-methallyl chain must be in close proximity to the Pd(II). As it is less sterically favorable for the O-methallyl chain to be close to the Pd(II) (conformer 16′), it will likely spend more time at a greater distance (conformer 17). This leaves the Pd(II) intermediate 16 more vulnerable to direct coupling. By replacing the hydrogen in the 3-position with another group, orientation of the O-methallyl chain toward the 3-position should become less favorable. This should result in the equilibrium being displaced toward 16′, therefore favoring cyclization.
To this end, the carboallylation conditions were applied to methyl-substituted potassium trifluoroborate 18 (Scheme 3). As predicted, the formation of direct allylation product 20 was greatly suppressed, with only a trace amount detected. The   During the development of their palladium-catalyzed cyclization of aryl halides, Zhou et al. also observed competition between cyclization and direct alkynylation. 5 The direct alkynylation process could be largely suppressed by using N-methylmethacrylamides, as shown in Figure 1b To examine the mechanism of the arylallylation process, a deuterium-labeling study was carried out. The heterocycle formation was carried out using aryltrifluoroborate 15 using dideuteroallyl bromide (Scheme  5). This resulted in the formation of dihydrobenzofuran 25 as a single deuterated isomer, and direct allylation products 26a and 26b as a 1:1 mixture of deuterated isomers (in 21 and 1% isolated yields, respectively). This is consistent with an isohypsic mechanism for the formation of 25 (Scheme 6). Beginning with transmetallation of boronic acid 10, palladium(II) intermediate 16 is formed. Carbopalladation (olefin insertion) forms the C−C bond of the dihydrobenzofuran ring, giving 27. A second carbopalladation can then occur, this time onto the allyl halide double bond giving palladium(II) intermediate 28. Finally, β-halide elimination gives the dihydrobenzofuran product 25 as a single deuterated isomer, and releases the palladium(II) catalyst.
However, an isohypsic mechanism is not consistent with the formation of a 1:1 mixture of deuterated isomers of the direct allylation products 26a/b. This suggests that the direct allylation products are formed via an alternative mechanism.
As the dihydrobenzofuran and direct allylation products appear to be formed by separate mechanisms, it should be possible to control, which product is formed by controlling the catalyst used. Having already shown the use of a palladium(II) catalyst to form a dihydrobenzofuran product from potassium trifluoroborate 15 (Table 5, entry 4), we chose to treat 15 with a Pd(0) catalyst. As predicted, exposure of 15 to Pd 2 dba 3 and allyl chloride resulted in the sole formation of direct coupling product 12 (Scheme 8). 11 Interestingly, use of several common oxidants, such as benzoquinone, DDQ, O 2 , or Cu(II), did not favor the isohypsic cyclization process.

CONCLUSIONS
We have developed a new Pd-catalyzed arylallylation reaction of alkenes. This reaction has been demonstrated in the formation of heterocycles, such as dihydrobenzofurans and oxindolines, results in formation of two new carbon−carbon bonds, and generates a quaternary carbon center. After elimination of the deboronation side product and suppression of the direct coupling product, the arylallylation reaction was shown to proceed through an isohypsic palladium(II)catalyzed mechanism. By controlling the reaction conditions, selective formation of either the cyclized or direct allylated product is possible.