Quaternary Stereocenters via an Enantioconvergent Catalytic S_N1 Reaction

Abstract

The unimolecular nucleophilic substitution (S_N1) mechanism figures prominently in every introductory organic chemistry course. In principle, stepwise displacement of a leaving group by a nucleophile via a carbocationic intermediate allows for the construction of highly congested carbon centers. However, the intrinsic instability and high reactivity of the carbocationic intermediates render it very difficult to control product distributions and stereoselectivity in reactions proceeding via S_N1 pathways. Here we report asymmetric catalysis of an S_N1-type reaction mechanism resulting in the enantioselective construction of quaternary stereocenters from racemic precursors. The new transformation relies on the synergistic action of a chiral hydrogen bond donor (HBD) catalyst with a strong Lewis acid promoter to mediate the formation of tertiary carbocationic intermediates at low temperature and achieve high levels of control over reaction enantioselectivity and product distribution. The work presented here provides a foundation for the enantioconvergent synthesis of other fully-substituted carbon stereocenters.

Quaternary stereogenic centers are important structural motifs in natural products and biologically-active compounds, conferring valuable structural, conformational and metabolic properties. Their construction has long been recognized as an important challenge to the field of chemical synthesis, and several distinct catalytic, enantioselective approaches have been developed in response.^1,2,3,4 Notable examples include cycloadditions,⁵ α and β-alkylation and arylation of carbonyls,^6–8 3,3′-additions,⁹ S_N2′ reactions,¹⁰ and Heck-type cross-couplings.¹¹ Each of these very powerful methods relies on enantiofacial addition across a prochiral substrate (Figure 1a) and therefore requires the preparation of stereochemically well-defined starting materials (such as trisubstituted olefins), and subsequent enantioselective bond formation.

Figure 1. Approaches to the enantiocontrolled construction of quaternary stereocenters.

a, Traditional methods for synthesis of quaternary stereocenter-containing molecules employ stereochemically defined prochiral substrates. b, S_N1 approach to the construction of quaternary stereocenters. c, Enantioselective allylation of propargyl acetates using chiral squaramide catalysts and TMSOTf promoter. TMS, trimethylsilyl; Np, naphthyl.

We envisioned that stepwise nucleophilic substitution reactions proceeding through prochiral carbocationic intermediates could provide a useful and complementary strategy for the enantioselective synthesis of compounds bearing quaternary stereocenters. Unlike the synthetic approaches noted above, quaternary-stereocenter construction via an S_N1-like pathway might be stereoablative¹² and thus could engage readily accessed racemic compounds as substrates (Figure 1b). While realization of this strategy would lift the requirement for stereocontrol in the synthesis of the substrate, it would also require overcoming several very significant challenges. The requisite catalytic system must (a) generate a reactive tertiary carbocationic intermediate, (b) minimize undesired elimination and rearrangement pathways, and (c) exert enantiocontrol in additions of a C-centered nucleophile to a high energy cationic intermediate. Indeed, despite the practical appeal of an enantioconvergent approach to the construction of quaternary stereocenters, only isolated examples have been reported to date.^9,13,14

Over the past decade, chiral, dual hydrogen-bond-donor (HBD) catalysts have been developed that promote enantioselective nucleophilic substitution and addition reactions via ion pair intermediates. These catalysts promote ion pair formation via direct anion abstraction¹⁵ or by substrate protonation with a co-catalytic Brønsted acid.¹⁶ Asymmetric induction is typically achieved from the resultant ion pair as a consequence of specific attractive non-covalent interactions between the corresponding cationic intermediate and the chiral HBD catalyst.^17,18,19 Reported examples have been limited to heteroatom-stabilized cations, due to the challenges in generating the requisite ion pair and suppressing elimination/rearrangement pathways. The ability of HBD catalysts to control enantioselective nucleophile addition into non-heteroatom stabilized carbocations has to our knowledge not been demonstrated.

Recently, our group discovered that chiral squaramide catalysts could be used in conjunction with Lewis acids such as trimethylsilyl trifluoromethanesulfonate (TMSOTf) to promote enantioselective reactions.²⁰ This dual-catalyst system was shown to promote the formation of oxocarbenium ions from dialkyl acetals – substrates that are unreactive under previously developed HBD-promoted reaction manifolds – while still engaging in attractive non-covalent interactions to achieve enantioinduction. We envisioned that the strong ionizing ability of this dual catalyst system could provide access to carbocationic intermediates lacking heteroatom stabilization, thus allowing us to examine whether small molecule H-bond donors can be used to promote productive, enantioselective reaction pathways from such high energy intermediates.

Reaction development

After an extensive evaluation of potential tertiary electrophile–C-centered nucleophile coupling partners, the reaction of propargyl acetate 2a with allyltrimethylsilane was identified as a useful model system to test this proposal (Figure 1c). In the absence of added HBD catalyst, the Lewis acid-promoted reaction affords a 1:1 mixture of desired product 3a to elimination product 4a (Figure 1c, entry 1). When readily accessed squaramide 1a (10 mol %) was added to the reaction, however, 3a was obtained in high yield (40:1, 3a:4a; Figure 1c, entry 2) and enantioselectivity (91% e.e.). Product ratio and enantioselectivity were strongly dependent on the nature of the HBD moiety: corresponding N,N-dimethylated squaramide (1b), thiourea (1c) and urea catalysts (1d) afforded 3a in low yield and e.e. (Figure 1c, entry 3–5). No reaction was observed with squaramide, thiourea, or urea HBD catalysts in the absence of TMSOTf.

A series of tertiary propargyl acetates was subsequently evaluated in order to probe the reaction scope and to generate preliminary information about the mechanism of the enantioselective substitution reaction (Figure 2a). Substrates bearing electron-donating (e.g. 2b and 2c) and electron-withdrawing substituents (2d) underwent allylation in high enantioselectivity (>90% e.e.) and product selectivity (>30:1 3:4). A linear correlation with a small negative slope (ρ⁺= −0.43, Figure 2b) was observed between Hammett substituent σ⁺ constants and log(e.r.) for substrates 2a–d. In contrast, a linear correlation with a very large negative slope (ρ⁺= −5.48, Figure 2c) was obtained from the corresponding plot of σ⁺ constants versus relative reaction rate (log(ν_X/ ν_H)) for the same substrates. The observation of a linear free-energy dependence (ρ⁺) of this magnitude provides direct evidence of positive charge accumulation in the rate-determining transition state, consistent with an S_N1-type ionization mechanism.²¹

Figure 2. Asymmetric allylation of propargyl acetates.

a, Substrate scope. Reactions were run on 0.6 mmol scale with 0.1 equiv. 1a, 1.0 equiv. TMSOTf, and 6.0 equiv. allyltrimethylsilane in 0.1M Et₂O at −78 °C for 24 h. ^a Reaction time was 4 h. ^b Reaction time was 14 d. ^c NMR yield. b, Hammett plot of σ⁺ values of substituents in 2a–d versus enantiomer ratios obtained in the formation of 3a–d. c, Hammett plot of σ⁺ values on substituents in 2a–d versus relative reaction rates determined for each substrate. d, Linear free energy plot of the calculated polarizability of the aromatic rings in 2a, 2e–g versus enantiomer ratios obtained in the formation of 3a, 3e–g. e, The absolute configuration of (–)-3b was determined by X-ray crystallography (structure shown), following derivatization to triazole 5b; the configuration of all other products was assigned by analogy. Conditions: (a) TBAF (2.0 equiv.), THF, r.t.; (b) 4-nitrobenzylbromide (1.1 equiv.), NaN₃ (1.1 equiv.), CuSO₄ (0.1 equiv.), sodium ascorbate (0.2 equiv.), tBuOH/H₂O (1:2), 50 °C; (c) HCl (3M in Et₂O).

Despite the very subtle dependence of e.e. on the electronic properties of substrate substituents noted above, reaction enantioselectivity was strongly responsive to changes in the expanse and position of the aryl moiety of the substrate. A linear correlation was observed between polarizability values calculated for the aryl substituent,²² and log(e.r.) of products 3a and 3e–g (Figure 2d), indicating that stabilizing aromatic interactions are likely to serve as a contributing factor in enantiodifferentiation.²³ Indeed, evidence for the existence of such stabilizing interactions could be gleaned from computational analysis of the putative complex between catalyst 1a and substrate 2a (Figure S9). Steric congestion near the reaction site also correlates with enantioselectivity. Thus, the o-tolyl-substituted derivative 2j underwent the allylation reaction to afford product with higher e.e. (82%) than the p- and m-substituted analogs 2h and 2i (66–67% e.e.). Similarly, ethyl-substituted product 3k was obtained in higher e.e. (94%) than methyl-substituted product 3b (c.f. 91% e.e.).

Substrates containing electron-rich heterocycles also underwent highly enantioselective substitution. Representative S– and O–heterocyclic substrates underwent reaction with allyltrimethylsilane to afford quaternary products (3l–n) in high yield and e.e., and with no detectable elimination by products. Following derivatization, the absolute stereochemistry of product 3b was determined by X-ray crystallography (Figure 2e).

Mechanistic studies

A mechanistic study of the reaction between a representative tertiary propargyl acetate substrate and allyltrimethylsilane promoted by squaramide 1a and TMSOTf was undertaken to glean insight into the underlying catalytic mechanism. The disappearance of 2b could be monitored over the entire course of the reaction using in situ infrared spectroscopy. Plots of the rate versus concentration data obtained from a “same excess” experiment revealed good graphical overlay (Figure 3a), demonstrating that no catalyst decomposition or product inhibition occurs over the course of the reactions.²⁴ The linearity of these plots further indicates that the reaction obeys a first-order rate dependence overall. Good overlay was also observed in the plots of the data obtained from “different-excess” experiments, revealing a first order rate dependence on [2b] and no rate dependence on the concentration of allyltrimethylsilane. These kinetic findings are consistent with a stepwise reaction mechanism where substrate C-O cleavage is turnover-limiting and nucleophile addition occurs in a post-turnover-limiting step (Figure 3b). Kinetic studies further revealed a sub-first-order dependence of reaction rate on [TMSOTf], and a first-order dependence of reaction rate on [1a] with a non-zero y-intercept. The kinetic dependence on the concentrations of both 1a and TMSOTf is consistent with pre-equilibrium formation of a resting state 1a•TMSOTf complex that reacts directly with substrate 2. The non-zero y-intercept is congruent with a competing background reaction observed in the absence of 1a (see Supplementary Information). The observation that optimal enantioselectivities are obtained under conditions where a background, uncatalyzed reaction is expected is intriguing, and the subject of continued study.

Figure 3. Kinetic data and catalytic cycle.

a, Reaction progress kinetic analysis of the reaction of 2b with allyltrimethylsilane. 0.047M; [allyltrimethylsilane]₀ = 0.195M; different excess: [2b]₀= 0.08M; [allyltrimethylsilane]₀ = 0.315M. b, Proposed catalytic mechanism for the enantioselective allylation of propargyl acetates.

With squaramide-TMSOTf-promoted formation of the carbocationic intermediate established as rate-limiting, a series of experiments was performed to interrogate the critical post rate-limiting steps. The formation of elimination byproduct was determined to be irreversible on the basis of a crossover experiment in which 1-naphthyl-substituted enyne 4a (0.25 equiv) was introduced to the reaction of 2-naphthyl-substituted acetate, 2f, under otherwise standard reaction conditions. This reaction afforded an 80% yield of 2-naphthyl-substituted product 3f, and alkene 4a was recovered in 97% yield with no trace of 1-naphthyl allylated product 3a detected (Figure 4a).

Figure 4. Mechanistic studies probing the post-rate-limiting steps of the allylation reaction.

a, Crossover experiment establishing irreversible formation of alkene byproduct. b, Partial reaction with scalemic 2g demonstrating that allylation proceeds via a stereoablative mechanism rather than by a dynamic kinetic resolution process. c, Predicted and measured ¹²C/¹³C kinetic isotope effects (KIEs).

To evaluate whether the reaction proceeds through an enantioselective or enantiospecific mechanism, the allylation was carried out by subjecting scalemic substrate (–)-2f (81% e.e.) to both enantiomers of the squaramide catalyst 1a (Figure 4b). After 1h reaction time, product 3f was obtained in 86% e.e. and 24% yield using (S)-1a; in the presence of (R)-1a, product 3f was obtained in similar yield but with opposite enantioselectivity (−85% e.e.). In both cases, the recovered substrate 2f was observed to have undergone only a small degree of epimerization, comparable to that observed when 2f was treated with TMSOTf and in the absence of squaramide catalyst. The results of these experiments are consistent with a stereoablative mechanism, i.e. an enantioselective process that proceeds through an achiral carbocationic intermediate. In contrast, a dynamic kinetic resolution pathway can be ruled out where 2f undergoes rapid racemization and one enantiomer preferentially undergoes stereospecific substitution.

Two limiting mechanistic possibilities were considered with regard to the enantiodetermining C–C bond-forming step: (a) irreversible nucleophile addition followed by rapid silyl elimination (Figure 4c, top), or (b) rapid and reversible nucleophile addition, followed by e.e.-determining silyl elimination (Figure 4c, bottom). These two scenarios are predicted to produce different carbon isotope effects at the allyl fragment. The carbon kinetic isotope effects (KIEs) were determined with natural abundance materials using Singleton’s NMR methodology (Figure 4c, see supplementary materials).²⁵ A significant primary KIE of 1.027 was observed at the position of bond formation (internal allylic methylene), while no KIE was observed at the terminal position. These results demonstrate that the first C–C bond-forming step is irreversible, therefore enantiodetermining.

Conclusion

The cooperative effect of chiral squaramides and TMSOTf was shown to generate tertiary carbocations lacking heteroatom stabilization from racemic precursors, control enantioselectivity in additions of a C-centered nucleophile, and attenuate undesired elimination pathways. The strategy outlined here may be generalizable to the construction of many types of highly congested stereogenic centers.