Selective acceleration of disfavored enolate addition reactions by anion–π interactions

The tortoise and the hare: anion–π interactions are reported to selectively accelerate the intrinsically disfavored addition of malonate half thioesters.

Cation-p interactions play a central role in molecular recognition, translocation and transformation. [1][2][3][4][5] Arguably the most spectacular manifestation of cation-p catalysis in biology is found in the biosynthesis of steroids, in which cascade cyclization occurs via carbocation hopping on a stabilizing cluster of p-basic amino acid residues (Fig. 1). 2 Cation-p interactions are also increasingly recognized in organocatalysis. 3-5 Contributions from the complementary but much younger 6 anion-p interactions 7 have been reported for anion binding [8][9][10][11] and transport. 11,12 In sharp contrast, explicit considerations of anion-p interactions in catalysis are extremely rare and very recent. [13][14][15] Looking for more signicant transformations that could benet from anion-p interactions, we considered malonate half thioester (MHT) 2, which is obtained by deprotonation of malonyl-CoA, a malonic acid half thioester (MAHT), under mildest conditions (Fig. 1). 16 Claisen condensation with acetyl-CoA 3 yields acetoacetyl-CoA as entry into both biosynthetic pathways, terpenoids and polyketides. Catalyzed by polyketide synthases, repeated decarboxylative enolate addition of the same substrate provides access to more than 10 000 natural products as important as fatty acids and lipids, macrolactones and higher aromatics. 16 In solution, MHTs have been shown to add to nitroolens, enones, aldehydes, imines or (thio)esters as in polyketide biosynthesis. [17][18][19][20][21][22] However, under unoptimized conditions, the non-productive decarboxylation to 3 dominates. Recent mechanistic studies indicate that for addition to occur, it should precede decarboxylation. 21,22 This should be possible with tautomers 4 or 5, whereas tautomer 2 should favor decarboxylation. Control over Fig. 1 In nature, carbocation chemistry in the biosynthesis of terpenes and steroids is accomplished with cation-p interactions (red circles indicate the position of p-basic amino-acid residues in the cation-p enzyme for substrate 1). The complementary enolate chemistry in polyketide biosynthesis and the beginning of both pathways fails in solution because decarboxylation of 2 (solid arrows) dominates over enolate addition (dashed arrows). In this report, selective acceleration of this disfavored but relevant process is achieved with anion-p interactions (blue background) and explained with the discrimination between non-planar tautomers (2) and planar tautomers (4/5; BH + : protonated base, E ¼ electrophilic carbon).
the selectivity between addition and decarboxylation thus calls for the discrimination between planar tautomers in which the negative charge is delocalized by resonance and tautomers in which planarity and resonance are disrupted by the tetrahedral sp 3 carbon between the two carbonyl groups. Anion recognition on p-acidic aromatic planes appeared just ideal to feel these subtle structural differences.
The addition of aromatic 18 and aliphatic MAHTs 6 and 7 to aromatic and aliphatic nitroolens 9-12,20 8 and 9 was selected to elaborate on this hypothesis (Fig. 2). In bifunctional anion-p catalysts, p-acidic 1,4,5,8-naphthalenediimide (NDI) derivatives [12][13][14][23][24][25] are appended to an amine base to provide stabilizing p surfaces for the enolate intermediates as soon as they are produced. All substrates and catalysts were synthesized from commercially available starting materials in a few steps (Scheme S1 †). 26 The reaction of MAHTs 6 and 7 with nitroolens 8 and 9 was followed by 1 H NMR spectroscopy, in which the evolution of the intrinsically disfavored addition products d (10) and the favored decarboxylation products f (3) was recorded with time against internal standards ( Fig. S1-S4 †). Results were quantied with h d/f values, that is the yield h d of the intrinsically disfavored divided by h f for the favored product (Table 1).
Catalyzed with TEA 11 at room temperature, h d/f ¼ 0.6 conrmed that the undesired decarboxylation is indeed favored under these conditions (Fig. 2, Table 1, entry 1). In comparison, the simplest possible anion-p catalyst, i.e., catalyst 12 composed of a p-acidic NDI surface next to a tertiary amine base, already gave rise to a slightly better h d/f ¼ 0.8 (Table 1, entry 2). The number of p-acidic surfaces in the catalyst was doubled next to increase the effective molarity of p-acidic surfaces or to even act from two sides on the reaction. The   resulting h d/f ¼ 0.9 demonstrated that with anion-p tweezer 13, addition became almost as good as decarboxylation (Table 1, entry 3). These encouraging results called for a systematic assessment of the contribution from anion-p interactions. The reversible oxidation of sulde donors to sulfoxide and sulfone acceptors has been introduced and validated previously as unique approach to vary p acidity with minimal structural changes. 18 Anion-p tweezer 14 with two suldes in the core of each NDI was prepared as a mixture of axial stereoisomers (Fig. 3b). The temperature-controlled stepwise sulde oxidation was unproblematic as long as the tertiary amine was protected rst against oxidation by protonation with TFA. Although insufficient, 7 the energy levels of the LUMOs are used as an approximative measure for p acidity. They decrease from E LUMO ¼ À3.9 eV for NDIs in 14 with two sulde donors to E LUMO ¼ À4.4 eV for NDIs in 15 with sulfoxide acceptors and E LUMO ¼ À4.6 eV for NDIs in 16 with sulfones. 12 All reactions proceeded to completion, with little formation of other side products ( Table 1). The nitroolen acceptor 8 was used in excess to maximize the probability of addition once the reactive enolate is formed on the p-acidic surface. For comparative evaluation, 20 mol% catalyst was used with regard to the MAHT substrate 6. With the best anion-p tweezer 16, selectivity ratios were with h d/f ¼ 6.0 nearly preserved at reduced catalyst loading (Table 1, entries 13 and 14). With 2 mol% 16, full conversion within 30 days at 5 C gave a turnover number TON ¼ 50 (Fig. S2 †).
The dependence of selectivity on p acidity, expressed as E LUMO of the catalysts, was close to exponential, 14b independent of temperature and substrates (Fig. 3a, C, A, ,). The compared to the perfect sulfur series 14-16 somewhat underperforming unsubstituted NDI tweezers 13 indicated the presence of minor, supportive as well as constant contributions from the ethyl sidechains to catalysis and thus conrmed the importance of the isostructural variation of p acidity provided by stepwise sulde oxidation in the series 14-16 ( Fig. 3a and 4). The selectivities obtained for controls 11 and 17 at different temperatures clustered below a virtual E LUMO ¼ À3.7 eV (Fig. 3a, O, P). This value corresponds to a nearly p-neutral NDI with two alkylamino donors in the core. 12,23 Selectivity values of controls coinciding with those extrapolated for p-neutral NDI surfaces provided corroborative support that anion-p Reactions with anion-p catalysts were slower than with TEA 11 but still much faster than without any amine catalyst. The initial velocities of product formation were used to determine activation energies E f a and E d a , that is the energy difference between ground state and transition state for the favored decarboxylation (f) and the disfavored addition (d, Fig. S5 †). Subtraction of activation energies of controls 11 or 17 from those of anion-p catalysts gave DE f a and DE d a (deceleration: DE a > 0, acceleration: DE a < 0). Positive DE f a and DE d a revealed that compared to control 11, anion-p tweezers 13-16 slowed down both processes (Table 1). Compared to the more revealing control 17, anion-p catalysts 13-16 always accelerated the disfavored (DE d a < 0) and mostly decelerated the favored process (DE f a > 0, Table 1). Most impressive the trends at low temperatures: without exception, acceleration of disfavored and deceleration of favored reaction both increased with increasing p acidity of anion-p tweezers 13-16 (Fig. 4a, Table 1, entries 10-13).
Selective acceleration of a disfavored reaction is given as a À DE f a < 0, valid for both deceleration or acceleration of the competing processes. Close to exponential 14b increase of the negative DDE d-f a with increasing p acidity of the catalyst was found, independent of conditions ( Fig. 4b, C (warm) vs. B (cold)), substrates (Fig. 4b, C (6) vs. , (7)) and controls (Fig. 4b, C (11) vs. X (17), Table 1). This consistent kinetic response to increasing p acidity supported that the inversion of selectivity indeed originates from anion-p interactions.
Selective deceleration of the favored decarboxylation and selective acceleration of the disfavored addition were both in agreement with the envisioned discrimination of differently planarized and delocalized tautomers by anion-p interactions (Fig. 1). It might be important to recall that direct experimental evidence for the ground-state stabilization of enolates on p-acidic surfaces is available from covalent model systems. 14 Transition-state destabilization for decarboxylation (by immobilizing the localized negative charge in tautomer 2 on the carboxylate oxygens on the p-acidic surface) could contribute as well. The same is true for transition-state stabilization for addition (by stabilizing the formation of the topologically matching nitronate 9-12 on the p-acidic surface). More explicit comments on mechanisms, applications and perspectives 26,13c would be premature. Such concluding remarks are also not needed to appreciate the main lesson learned from this study: selective "tortoise-and-hare catalysis" 27 of enolate chemistry provides experimental support that anion-p catalysis 13 not only exists but also matters. Fig. 4 (a) Dependence of the changes in activation energy DE a for substrate 6 for the favored decarboxylation (DE f a , B) and the disfavored addition (to 8, DE d a , C) on the p acidity of anion-p tweezers 13-16 (E LUMO ), relative to control 17, at 5 C, with exponential curve fit ( Table 1, entries 10-13). (b) Selective acceleration of a disfavoured reaction: dependence of DDE d-f a , i.e., DE d a À DE f a , for 6 on the p acidity of 13-16 (E LUMO ) compared to control 11 (C, A, ,) or 17 (X) at RT (C, X, ,) and 5 C (A) for 6 (C, X, A) and 7 (,) with 8, with exponential curve fit (compare Table 1).