Anion–π catalysis: bicyclic products with four contiguous stereogenic centers from otherwise elusive diastereospecific domino reactions on π-acidic surfaces

Delocalized over aromatic planes, anion–π interactions emerge as best to stabilize long-distance charge displacements in domino reactions of highest sophistication.

enantioselectivity with increasing rates, i.e., asymmetric transition-state stabilization in the presence of p-acidic surfaces and inhibition with the anion selectivity sequence NO 3 À > Br À > BF 4 À > PF 6 À .
Contributions of anion-p interactions 1,2 to catalysis were rst demonstrated explicitly in 2013. 3 Since then, anion-p catalysis has been explored with enolate, 4,5 enamine, 6 iminium, 7 transamination 8 and oxocarbenium 9 chemistry, and the rst anion-p enzyme has been created. 5 The idea of stabilizing anionic intermediates and transition states on p-acidic surfaces is a new fundamental concept. Delocalized over large aromatic planes, anion-p interactions appear particularly advantageous to stabilize extensive charge displacements over long distances. For the conventional cation-p catalysis, this advantage is best illustrated with the stabilization of carbocations moving along the emerging rings during the cyclization of terpenes into steroids. 10 In anionic tandem, domino or cascade reactions, the key intermediates in need of stabilization are not carbocations but enolates, nitronates, and so on. The most sophisticated domino reaction catalyzed so far with anion-p interactions is the stereoselective formation of a cyclohexane ring from achiral starting materials. 7 From there, a continuing increase in sophistication of long-distance cascade charge displacements on p-acidic surfaces naturally leads to bicyclic products on the one hand and quaternary stereogenic centers on the other. For this purpose, we focused rst on the addition of cyclohexanedione 1 to nitroolen 2 (Fig. 1). In the presence of a base, they engage in a domino Michael-Henry reaction to afford bicyclo[3.2.1]octan-8-one 3, which is a bicyclic product with four chiral centers made from achiral substrates. [11][12][13] The rst step is the Michael addition of the conjugate enolate base of 1 to acceptor 2 (see transition state TS1, Fig. 1). The resulting nitronate engages in an intramolecular Henry reaction to close the second carbocycle (TS2, Fig. 1). This reaction was attractive for anion-p catalysis because stabilization of the anionic enolate and nitronate intermediates on p-acidic surfaces was conceivable, and the diastereoselectivity reported in the literature for metal-free organocatalysts (up to 12 : 1) appeared improvable. [11][12][13] One report highlights that poor 1 : 3 dr originates from epimerization between 3 and 3d in the presence of base catalysts. 11 Here, we report that domino catalysis on pacidic surfaces provides access to diastereospecicity, i.e., the exclusive formation of one diastereomer. This breakthrough with the most sophisticated tandem process realized so far is achieved with three different functional systems, i.e., new anion-p cinchona fusion catalysts, anion-p enzymes and achiral anion-p chirality enhancers. Decisive contributions from anion-p interactions could be deduced from the observed increase in rate and stereoselectivity upon attachment of pacidic surface, and selective inhibition with anions.
To elaborate on the formation of bicyclic products from achiral diketones on p-acidic surfaces, catalysts and controls 4-14 were prepared (Fig. 2). Most were accessible following reported procedures; details can be found in the ESI (Schemes S1-S5 †). Anion-p catalyst 4 has been designed around the p-acidic surface of a naphthalenediimide (NDI). 2 NDIs offer a privileged platform in anion-p catalysis because their intrinsic quadrupole moment perpendicular to the p surface is very high and can be easily modulated with substituents in the core. 2-8 In anion-p catalyst 4, this p surface was connected to a tertiary amine catalyst via a xed Leonard turn. 4 These turns have been introduced to assure that the reactions really occur on the p surfaces and benet best from anion-p interactions, even when they are not particularly strong. 4 The imide on the other side of the p-acidic surface continues with a simple solubilizing group.
In the presence of 10 mol% of anion-p catalyst, the formation of bicyclo[3.2.1]octan-8-one 3 occurred within 1-3 days at ambient temperature. The known absolute conguration of product 3e obtained with quinine 15 allowed us to assign the absolute conguration of enantiomer 3 obtained with 4-12 ( Fig. 2d). 12 Solvent screening with 4 gave best results in C 6 F 6 , i.e. 10 : 1 dr and 80% ee (Table 1, entries [12][13][14][15]. With 20% C 6 D 6 , a slight increase to 13 : 1 dr coincided with a slight decrease to 77% ee (Table 1, entry 7). Further decreasing p acidity of the solvent gradually decreased stereoselectivity down to 6 : 1 dr and 66% ee in C 6 D 6 (Table 1, entry 4). These trends were interesting because they supported contributions from synergistic anion-p interactions with polarization of the NDI plane induced by the solvent. 14 From C 6 F 6 , enantioselectivities could be further increased with co-solvents, best with 88% ee in C 6 F 6 / CDCl 3 2 : 1, but diastereoselectivity dropped to 7 : 1 dr under these conditions (Table 1, entry 9).
In C 6 F 6 , catalyst loadings could be reduced to 2.5 mol% without signicant losses in stereoselectivity (Table 1, entries 12-14). At 5 mol% in C 6 F 6 , reduction of the temperature to 5 C further increased diastereoselectivity to 13 : 1 dr, whereas enantioselectivity did not change signicantly (Table 1, entry 15).  Variation of the core substituents did not affect the activity of catalysts 4-7 signicantly, also because the p acidity of the p surface was not much changed (Table 1, entries 13 and 16-18). 15 The best activity was obtained for 5 with phenylsuldes in the core, i.e., 83% ee and 13 : 1 dr for 5 mol% in C 6 F 6 (Table 1, entry 16). The 90% ee obtained at maximal p acidity without core substituents in 7 was accompanied by a decrease in diastereoselectivity to 7 : 1 dr (Table 1, entry 18). Similarly insignicant were variations of the second imide substituent, i.e. R 2 opposite to the Leonard-turned amine in catalysts 8-10 (Table 1, entries 19, 21 and 22). This trend suggested that these anion-p catalysts are formally bifunctional, but not trifunctional (Fig. 1).
Cinchona alkaloids are most popular in amine-based asymmetric organocatalysis. 11,12,16 In the newly designed catalyst 13, this bicyclic amine was attached to a p-acidic surface via a Leonard turn, similar to the one introduced in catalysts 4-10 (Fig. 2). The synthesis of the new anion-p cinchona fusion catalyst 13 was very straightforward (Scheme S4 †). In C 6 F 6 / CDCl 3 4 : 1, 5 mol% of cinchona catalyst 13 produced bicycle 3e in 89% yield with À94% ee and only one detectable diastereomer, i.e., >20 : 1 dr (Table 1, entry 23). This shi from diastereoselectivity to diastereospecicity is unprecedented for this domino process; previous records in the literature, achieved with cinchona catalysts interfaced with conventional thioureas, stopped at 12 : 1 dr. 12 The diastereospecicity obtained with cinchona catalyst 13 suggested that on p-acidic surfaces, either the epimerization between 3e and 3ed is suppressed or the protonation of the nitronate intermediate is stereospecic.
To position anion-p catalysts within the chiral space of proteins, anion-p catalyst 4 has been coupled with a biotin. 5 Binding of the resulting conjugate 16 to streptavidin then afforded an articial anion-p enzyme, which operates with an essentially unknown interaction to biological enzymes (Fig. 3a). 5 Interfacing of anion-p catalysts with proteins is attractive because access to mutant screening allows performance to be readily optimized. This screening approach has previously afforded anion-p enzymes that catalyze, at pH 3.0, the addition of malonic acid half thioesters to enolate acceptors with 95% ee and unprecedented chemospecicity with regard to the intrinsically favored decarboxylation. 5 A focused mutant screening for the domino reaction to bicycle 3 in the presence of 16 gave best results for S112W at pH 6.5 (À76% ee, Table 1, entry 39). Mutant S112Y, the best for the addition of malonic acid half thioesters, was with À53% ee slightly better than the wild-type (WT) protein with À45% ee (Table 1, entries 35 and 38). K121 was conrmed as essential, presumably to keep the tertiary amine base in 16 from protonation under experimental conditions (Table 1, entries 36 and 37). In the presence of nitrate, the enantioselectivity decreased, with IC 50 ¼ 0.34 M (Fig. 3b, Table  1, entry 40). This nding provided experimental support for operational anion-p interactions, i.e. the existence of anion-p enzymes. Interestingly, all anion-p enzymes obtained with 16 gave enantiomer 3e with a maximum ee of À76% as the main product in nearly neutral water/MeCN 2 : 1, whereas the protein-free analog 4 gave the opposite enantiomer 3 with a maximum ee of 88% as the main product in C 6 F 6 /CDCl 3 4 : 1. It is perhaps this inversion of the intrinsic enantioselectivity induced by the xed Leonard turn derived from 1R,2R-diaminocyclohexane in both 4 and 16 that hindered access to higher ee's with anion-p enzymes.
In summary, this study drives the development of anion-p catalysts to unprecedented sophistication with regard to anionic domino reactions that take place on p-acidic surfaces. The most important ndings are anion-p cinchona fusion catalysts that clearly exceed the performance of conventional metal-free organocatalysts for the rst time, the rst example for diastereospecicity, anion-p enzymes that operate in neutral water and the discovery of achiral tetraalkylammonium salts as supramolecular chirality enhancers. The discovery of the best anion-p catalysts with the most sophisticated domino reaction supports the important expectation that anion-p interactions, delocalized over large aromatic planes, will be most advantageous in stabilizing extensive long-distance charge displacements during multiple coupled anionic intermediates and transition states. Long-term perspectives include the discovery of otherwise inaccessible reactions with anion-p catalysis, the integration into more complex systems, 5,17 and the introduction of other unorthodox interactions to catalysis. 3,18 Fig. 3 a) The concept of anion-p enzymes, with indication of the structure of the anion-p biotin conjugate 16 interface used with streptavidin mutants and the position of the mutated key residues from monomers A and B of the streptavidin tetramer, and (b) the dependence of the ee of 3e produced by the S112W mutant on the concentration of NO 3 -(NaNO 3 ).