Palladium-Catalyzed Decarboxylative Asymmetric Allylic Alkylation of Thietane 1,1-Dioxides

A palladium-catalyzed decarboxylative asymmetric allylic alkylation of thietane 1,1-dioxides via linear enolate intermediates from racemic starting materials has been developed. This process installs an α-sulfonyl tetrasubstituted stereogenic center with high enantioselectivity. The potential to transform the alkylated products to novel types of enantioenriched spirocycles for medicinal chemistry applications has also been demonstrated.

F our-membered-ring-containing spirocycles have become particularly attractive building blocks in drug discovery, 1 with much attention placed on the development of synthetic routes to achiral (1, Scheme 1A), 2 as well as chiral but racemic (2) spirocycles. 3 In contrast, chiral, enantiopure analogues 3 are much less common. 4 In particular, there are no examples of enantiopure thietane 1,1-dioxide containing spirocycles 4.
To access tetrasubstituted carbon centers in enantioenriched form, we sought to utilize the palladium-catalyzed decarbox-ylative asymmetric allylic alkylation (Pd-DAAA) reaction, 5 most frequently employed in the asymmetric alkylation of cyclic enolates. 6 However, the DAAA reaction of linear enolates has been less developed due to the need for stereoselective enolization of the carbonyl substrates in order to achieve high levels of asymmetric induction in the alkylation step. 7 The Stoltz group discovered that the Pd-DAAA reaction of linear enol carbonate 5 gives 6 with high enantioselectivity irrespective of the ratio of E/Z enol carbonates 5 due to a palladium-mediated interconversion of the intermediate enolates prior to alkylation (Scheme 1B). 7h In contrast to enolates, the asymmetric allylic alkylation of α-anions of sulfones is more challenging. 8 Tunge and coworkers developed an enantiospecific, stereoretentive decarboxylative allylic alkylation of linear sulfones 7 to 8 (Scheme 1C). 8b Their study revealed that allylic alkylation occurred faster than racemization of the α-sulfonyl anion, retaining the stereochemical information in the process. The enantioselective allylic alkylation of racemic α-sulfonyl nucleophiles remains elusive. 9 To enable an enantioconvergent alkylation of racemic sulfones, we incorporated a carbonyl group in thietane 1,1dioxide 9 as a means of simultaneously stabilizing the αsulfonyl anion and ensuring complete stereoablation via planar enolate 10 (Scheme 1D). However, as decarboxylation would likely lead to a mixture of E/Z enolates 10, a palladiummediated interconversion of the enolates would be required in order to obtain 11 with high ee. Herein we report the first palladium-catalyzed asymmetric allylic alkylation of thietane 1,1-dioxides to generate α-sulfonyl stereogenic tetrasubstituted carbon centers in 11 from racemic starting materials without the need for geometrically pure, preformed enol carbonate precursors. We illustrate the utility of these products in the synthesis of the novel, enantioenriched thietane 1,1-dioxide containing spirocycle 12.
Our studies began with a three-step synthesis of precursors 16 and 17 (Scheme 2). The oxidation of commercially available thietane (13) to thietane 1,1-dioxide (14) and allyl ester installation gave 15, which was derivatized in a divergent manner to substrates 16 and 17, bearing ketone and ester substituents, respectively.
The development of the Pd-DAAA process was undertaken using ketone substrate 16a (Table 1; see the Supporting Information for the full optimization study). Initial experiments to identify a suitable ligand for the enantioselective conversion of 16a to 18a revealed that the use of PHOX L1, as well as DACH phenyl and naphthyl Trost ligands L2 and L3, resulted in poor ee in 1,4-dioxane as the solvent (8%, 34% and 24% ee, respectively, entries 1−3). In contrast, (S,S)-ANDEN Trost ligand L4 gave a high ee of 83% of 18a (entry 4), despite the acyclic nature of the enolate intermediate. It was found that a reaction of lower concentration (0.04 M, entry 4) yielded 18a with higher ee in comparison to those with greater concentrations (0.1 and 0.2 M, entries 5 and 6, respectively). Reactions in acetonitrile and toluene were poorly selective (entries 7 and 8). 1,4-Dioxane was also superior to other ethereal solvents (entries 9 and 10), presumably due to its ability to better stabilize caged ion pairs. 10 Despite obtaining an excellent yield (98%) of 18a in CH 2 Cl 2 (entry 11), the enantioselectivity was also lower in comparison to 1,4-dioxane. Attempts to lower the temperature to 0°C using a 3/1 1,4dioxane/CH 2 Cl 2 mixture resulted in no improvement in ee (entry 12), with the reaction in 1,4-dioxane at room temperature being optimal (entry 4).
With the optimal conditions identified, the substrate scope was investigated by subjecting racemic precursors 16 and 17 to the catalytic reaction (Scheme 3). We discovered that electronics had little effect on the enantioselectivity of the reaction, with all substituted phenyl aromatic ketones giving ee values of products 18a−f of 83−86%. In spite of the increased size of the o-toluoyl ketone substituent in 18b, the yield and ee remained high. p-Fluoro-and bromo-substituted phenyl substituents were tolerated, with no oxidative addition occurring at the C−Br bond. A slightly lower ee of 79% was obtained for methyl ester containing product 18g. Aromatic heterocycles, such as furyl-and pyridyl-containing 18h,i, were obtained with 81% and 72% ee, respectively. Substitution on the allyl group at the internal or terminal position (18j−m) necessitated a higher catalyst loading due to the lower reactivity of these substrates, and the ee also dropped significantly to 52−55%. Prenylated substrate 18n did not undergo the reaction, presumably due to large steric hindrance at the allyl functionality for the initial oxidative addition step. Alkyl-substituted ketones were isolated with consistently high ee values of >90% in the case of large substituents, such as adamantyl (18o), tert-butyl (18p), cyclohexyl (18q), and isopropyl (18r). However, as the steric bulk of the substituent decreased, the enantioselectivity of the reaction fell: the ee values of 18s,t were 81% and 69%, respectively, whereas product 18u bearing a methyl ketone substituent was isolated with a much lower 39% ee. Using the same reaction conditions, ester-substituted products were all obtained with high enantioselectivity, including phenyl aromatic esters 19a−c, tert-butyl ester 19d, and methyl ester 19e. X-ray crystal structures of 19a,b confirmed the absolute stereochemical configuration of the major enantiomer of product using (S,S)-ANDEN phenyl Trost ligand L4. 11 Substrates with an amide as the stabilizing group did not undergo decarboxylation under the optimized conditions.
Given that the alkene geometry of acyclic enolates can affect the stereochemical outcome of the allylic alkylation reaction, 7d,h−j we sought to explore the importance of the enolate geometry on both the sense and magnitude of enantioinduction. In this context, the E-and Z-enol carbonates 20 were  . Both isomers of 20 afforded (S)-18p as the major enantiomer in 76% ee from (E)-20 and 88% ee from (Z)-20 with (S,S)-L4 as the chiral ligand. In the case of (E)-20, the level of enantioselectivity was slightly lower than that when allyl ester 16p was used as the substrate (90% ee of 18p). We therefore postulate that a palladium-mediated interconversion of E-and Z-enolates occurs and that alkylation of the Z-enolate results in the formation of 18p. (For a rationale of the origins of stereocontrol using the Trost "wall-and-flap" model, see the Supporting Information.) To gain further insight into the mechanism of this reaction, an enolate crossover experiment of 16a and deuterium-labeled [D]-16c was performed (Scheme 5A). The isolated product mixture comprised all four crossover compounds 18a, [D]-18a, 18c, and [D]-18c, as confirmed by high-resolution mass spectrometry. The complete scrambling of enolates suggests that the ion pairs generated in this reaction undergo fast ion exchange. 7d To test whether an enolate as part of an ion pair is a long-lived intermediate in the reaction, a water additive was used (Scheme 5B). We expected water to quench a free enolate to at least some extent if such a species was present in the reaction and significantly affect the yield, and potentially ee, of product 18a. However, neither the yield nor the enantioselectivity of the reaction was affected even when up to 20 equiv of water was added (see the Supporting Information for further details), indicating that a free enolate is an unlikely species in the reaction.
Finally, to elucidate whether an inner-or outer-sphere enolate alkylation operates, substituted allylic electrophile cis-21 was prepared (Scheme 5C). In the case of an outer-sphere alkylation, net retention of the allylic center would be expected in 22. Alternatively, the inner-sphere mechanism would result in net stereochemical inversion in 23. Unfortunately, 21 failed to undergo the desired alkylation due to the sterically encumbered nature of the allylic electrophile (see the Supporting Information for further details).
Using this information, a catalytic cycle for the Pd-DAAA reaction of thietane 1,1-dioxide is proposed (Scheme 6).  Following ionization of 16, palladium-carboxylate ion pair 24 is formed. We believe it is at this stage that ion crossover occurs. Given that the reaction is unaffected by water, it is likely that a free enolate is not formed due to the slow decarboxylation of 24. 12 This decarboxylation step is assisted by palladium (25), 13 which facilitates the requisite E/Z enolate equilibrium between 26a and 26b via a carbon-bound palladium enolate to afford enantioenriched 18 and release the palladium(0) catalyst. The enantioenriched tetrasubstituted thietane 1,1-dioxides obtained by the Pd-DAAA methodology are excellent building blocks for further derivatization into novel spirocycles (Scheme 7). In this context, the key allylic alkylation process of 17d was scaled up, furnishing 19d in 86% yield and 96% ee on a 5 g scale. The alkene in 19d was subjected to a hydroboration, oxidation, and reductive amination sequence, and removal of the tert-butyl group afforded amino acid 27. Subsequent lactamization, amide reduction, and deprotection furnished spirocycle 12. Finally, to exemplify the utility of spirocycle 12 in the generation of compound libraries, the amine in 12 successfully underwent reductive amination, amide bond formation, and Buchwald−Hartwig processes, efficiently furnishing products 28a−c.
In conclusion, we have developed the first enantioconvergent palladium-catalyzed DAAA reaction of thietane 1,1dioxides, in which a carbonyl substituent enables the key stereoablation required to produce enantioenriched alkylated products from racemic precursors. In spite of the likely formation of both E-and Z-enolates in the reaction, a palladium-mediated enolate interconversion is thought to occur, resulting in high levels of enantioselectivity. The synthetic utility of these products has been demonstrated by their expedient derivatization into a novel, enantioenriched thietane 1,1-dioxide containing spirocycle as a high-value sp 3