Rhodium-Catalyzed Allylic Amination for the Enantioselective Synthesis of Tertiary β-Fluoroamines

Abstract

The utilization of β-fluoroamines as pharmaceutical components for drug development has attracted a considerable amount of interest. However, direct access to tertiary β-fluoroamines is challenging. We herein report the rhodium-catalyzed asymmetric amination of tertiary allylic trichloroacetimidates with anilines and cyclic aliphatic amines to access tertiary β-fluoroamines, where the α-carbon atom is bonded to four different substituents, in good yield with high levels of enantioselectivity.

Graphical Abstract

Amines play a crucial role in pharmaceuticals and biochemical processes.¹ However, the inherent basicity of amino functionalities in small-molecule drugs may hinder these compounds from crossing the cell membrane.² Moreover, basic amines in drug formulations have been linked with several adverse effects, including hERG binding,³ cardiotoxicities,⁴ and phospholipidosis,⁵ thereby impacting their suitability for biomedical applications. Consequently, tuning the basicity of amines has emerged as a critical factor in drug design.⁶ Fluorine has been widely employed as a synthetic tool to modulate the basicity of amines.⁷ The pK_a, modulating effect by fluorine has accounted for improved biomedicinal properties (Scheme 1A).⁸ Overall, β-fluorinated amines have garnered pharmaceutical interest as potential drug candidates (Scheme 1B), with ~150 β-fluoroamine drug candidates currently in phase II and III clinical trials.⁹ The increasing demand for β-fluoroamine motifs has encouraged the continuous development of various synthetic methodologies. Nonetheless, the direct synthesis of β-fluoroamines remains challenging, particularly when the carbon atom bearing the amine is a chiral quaternary carbon bonded to four different substituents (Scheme 1C). The complexities associated with their synthesis are evident in multiple patents,^10–12 which propose accessing the tertiary β-fluoroamine as a racemic mixture and then performing asymmetric separation.¹³ The critical challenge in achieving direct asymmetric access to tertiary β-fluoroamines is the inherent difficulty in accessing the enantiopure C–N quaternary center.^14–16 Over the years, several strategies have been developed to address this issue, including cyclization reactions,¹⁷ α-functionalization of imines,¹⁸ and rearrangement reactions.³² These methods have enabled the direct synthesis of the C–N quaternary center with various degrees of success and limitations.^19,20 Transition metal catalysis has shown the most promise among these strategies.^21,22 In particular, transition metal-catalyzed asymmetric allylic substitution has proven to be a powerful technique for introducing the chiral C–N quaternary center.^23–26 Our group has reported an efficient strategy for the enantioselective construction of the C–N quaternary center using rhodium-catalyzed allylic amination of primary²⁷ and secondary amines²⁸ (Scheme 1D). Building on our previous work, we postulated that replacing the methyl group in allylic trichloroacetimidate 1a with a fluoromethyl group 1 could produce the challenging tertiary β-fluoroamine (Scheme 1E). In this context, we herein report the development of chiral rhodium-catalyzed allylic amination to construct tertiary β-fluoroamines asymmetrically.

Scheme 1. β-Fluoroamine in Bioactive Molecules and Experimental Design^a.

^a(A) Effect of a β-fluorine atom on the pK_a of tertiary amine pharmaceutical motifs. (B) Examples of bioactive molecules containing tertiary β-fluoroamines. (C) Classification of β-fluoroamines. (D) Previous work, which was rhodium-catalyzed allylic amination. (E) Design and development of the challenging allylic amination of fluoromethyl-substituted tertiary trichloroacetimidate.

One considerable challenge in allylic amination of tertiary substrates is achieving excellent regioselective control.^29,30 The presence of fluorine could diminish the regioselectivity via a secondary interaction at the coordination complex.^31,32 A key concern is the potential for such an interaction to disrupt the necessary bidentate coordination of the olefin and imine critical for ionization and selectivity (Scheme 1D).^44,45 To evaluate this possibility, we subjected allylic trichloroacetimidate 1 and 4-methoxyaniline 2 to regioselective amination conditions previously developed by our group (Table 1A).³³ The amination reaction proceeded to afford the desired tertiary β-fluoro amine 4 in 55% yield with an excellent linear:branch ratio (>1:99 l:b) (Table 1A). Control experiments showed that the conversion of the allylic trichloroacetimidate to the β-fluoroamines did not occur without the catalyst (see page S6 of the Supporting Information). A comparison between observed β-fluoroamine product 4 and previous amination product 4a (Table 1A) revealed that the reaction required a longer time, resulting in a slightly lower yield (from 72% to 55%).

Table 1.

Optimization of the Reaction Conditions^a

entry	Rh loading (mol %)	2 (equiv)	solvent	time (h)	yield (%)	ee (%)
1	5	1.5	THF	24	39	73
2	5	1.5	toluene	24	59	85
3	5	1.5	toluene	6	47	85
4	5	3.0	toluene	24	72	73
5	10	1.5	toluene	24	77	92

^aReaction conditions: tertiary allylic trichloroacetimidate 1a or 1 (1 equiv), amine 2 (3 equiv), THF (0.2 M). Isolated yield. (B) All amination reactions were conducted at 0.1 M with 1. Isolated yields are reported; the enantiomeric excess (ee) was determined using chiral HPLC.

Nevertheless, encouraged by the success of the regioselective results, we shifted our focus to the more challenging asymmetric amination of tertiary β-fluoroamines (Table 1B). Hayashi’s chiral diene ligand³⁴ was selected for our reaction because the previous asymmetric amination of the tertiary trichloroacetimidate substrate has reported high levels of enantioselectivity using this chiral ligand.^27,28 Accordingly, the asymmetric reaction of 4-methoxyaniline 2 using the chiral diene-ligated rhodium catalyst provided the desired fluorinated amine product 4 in 39% yield with moderate enantioselectivity [73% ee (Table 1B)]. Several solvents were screened (see page S6), and toluene gave the best yield (59%) and selectivity (85% ee, entry 2). Optimization conditions revealed that the yield and enantioselectivity of 4 were dependent on the catalyst loading, the number of amine equivalents, and reaction time (page S6). Notably, fluorine may affect enantioselectivity by hindering the formation of the π-allyl intermediate. Increasing catalyst loading facilitates ionization, consequently increasing the reactive π-allyl population and improving enantioselectivity.

With the optimized condition for 4 in hand [77%, 92% ee (Table 1B, entry 5)], we explored the scope of amines. Reactions involving various anilines proved to be effective, forming tertiary β-fluoroamines 4–19 in good yield with high enantioselectivity (Scheme 2A). Both electron-rich and electron-withdrawing anilines demonstrated suitability for the amination reaction. The reaction of pharmaceutically relevant 4-bromo-aniline and 4-aminophenylboronic acid pinacol ester provided tertiary β-fluoroamines 5 (41%, 96% ee) and 13 (70%, 96% ee), respectively, which are suitable for further postmodification via cross-coupling.^35,36 In addition to tertiary allylic trichloroacetimidate 1, the functional group tolerance of the fluoromethyl trichloroacetimidates was investigated. Both electron-deficient and electron-rich benzyl ether derivatives yielded products 10 and 11, respectively, in excellent enantioselectivity (96% ee), comparable to that of their benzyl ether derivative 9 (97% ee). Substituting the benzyl ether substrate with a naphthyl ether, forming 12 and 19, was also feasible. The formation of fluoromethyl products exhibited a high degree of asymmetric induction; however, the relatively modest yields can be attributed to the low reactivity of the tertiary allylic trichloroacetimidate substrates. Notably, unreacted starting materials persisted in the reaction mixtures even after 24 h, indicating incomplete conversion.

Scheme 2. Substrate Scope of Amines and Tertiary Allylic Trichloroacetimidates^a.

^a Reaction conditions: amine (1.5 equiv), trichloroacetimidate (1 equiv), and 10 mol % [RhCl(S,S)-L]₂ in 0.1 M toluene under a nitrogen atmosphere at 25 °C for 24 h. Isolated yields were calculated. The ee values were determined by chiral HPLC.

Next, we explored the substrate scope of secondary aliphatic amines. Under optimized conditions, the reactions provided β-fluoromethyl tertiary amines 20–27 in good yields and enantioselectivities (Scheme 2B). For instance, the use of indoline furnished 20 in 59% yield and 91% ee. Interestingly, enantiopure (R)-2-methyl indoline produced product 21 in comparable yield and excellent diastereoselectivity (dr = 17:1). The reaction between primary aliphatic amines and the trichloroacetimidate electrophile yielded no product.

To demonstrate the synthetic utility of our asymmetric access to tertiary β-fluoroamines, we also assessed the aminations using derivatives of pharmaceutical active amines. Tertiary β-fluoroamine products 28–32 afforded synthetically useful yields and enantioselectivities (Scheme 1C). In addition, the rhodium-catalyzed tertiary amination process demonstrated its applicability for the scalable synthesis of tertiary β-fluoroamines 4, utilizing 1 mmol of starting material 1. This approach afforded 4 in 59% yield and 87% ee (see pages S8 and S79).

Also, we compared the reactivity and enantioselectivity of the methyl-substituted trichloroacetimidate substrates with their fluoromethyl derivatives to investigate the impact of the fluorine atom on some selected substrates (Scheme 2). Notably, the enantioselectivities of the tertiary β-fluoroamines were comparable to those of previously reported tertiary amine products.^27,28 However, we did observe a decrease in the yields of the fluoroamine products. For instance, while fluorinated aniline 13 (Scheme 2A) was obtained with 70% yield and 96% ee, its methyl derivative 13a was formed with 86% yield and 91% ee. A similar trend was observed with other fluorinated products and their methyl analogues (10a, 15a, 21a, and 25a). These results suggest that the fluoromethyl group impacts the reactivity of the reaction (Scheme 2A,B) without any significant impact on the enantioselectivity.

To better understand the mechanism of the allylic amination, we isolated unreacted starting material 1 and determined its enantiomeric purity (see page S22). The findings revealed that the enantiomeric excess of 1 showed a modest increment (30% ee) after reaction for 24 h. Undoubtedly, this slow reactivity of tertiary fluoromethyl trichloroacetimidate and its modest enantioenrichment deviate from the results of our previous study.²⁸ Despite the observed modest increase in enantiomeric purity, this outcome does not suggest a kinetic resolution, as a corresponding lower yield should be expected, which was not evident in our experimental results.^37,38 We hypothesize that the observed enantioenrichment is a result of the unequal rates of ionization between the two enantiomers of racemic starting material 1.^39,40 However, we were unable to conduct kinetic studies of individual enantiopure trichloroacetimidates²⁸ to confirm this hypothesis due to difficulties in preparing enantioenriched starting material 1.⁴¹ In light of the similarity between the current amination of the fluoromethyl-substituted substrates and the established methyl-substituted tertiary allylic substrates,²⁸ we proposed a similar DYKAT mechanistic pathway for the formation of the tertiary β-fluoroamines (Scheme 1D).^42,43

The reactivity comparison between fluoromethyl-substituted trichloroacetimidate 1 and methyl-substituted derivative 1a (Table 1) raises questions. Overall, we observed that fluoromethyl substrate 1 exhibits lower reaction yields and longer reaction times and requires higher catalytic loading compared to that of methyl substrate 1a. We hypothesize that the reduced reactivity of the substrate with fluoromethyl substitution may be due to the limited presence of reactive π-allyl intermediates (Scheme 1D). Our previous study has shown that an electron-deficient group around the π-complex makes it difficult to form the π-allyl intermediates.⁴⁴ To elucidate the impact of the fluorine atom on the reactivity of the trichloroacetimidate, we utilized density functional theory (DFT) calculations to investigate the coordination mode and the crucial ionization transition state that leads to the reactive π-allyl intermediates. To streamline the energy analysis and minimize computational expenses, we applied the computational method previously documented by our group for the amination of the methyl-substituted tertiary allylic trichloroacetimidate.²⁸ The optimized energies were computed by substituting the methyl substituent of (S)-E1 and (R)-E1 with a fluoromethyl group, (S)-E1-F and (R)-E1-F, respectively, in the model structure (Figure 1). Extensive conformational analysis of the fluoromethyl trichloroacetimidate revealed a sustained preference for the bidentate coordination as well as an extended intramolecular con-covalent interaction between the highly polarized F–C–H group and the β-oxygen atom⁴⁵ (see pages S24–S27).

Figure 1.

Reactive Gibbs free energies (kilocalories per mole) for ionization of the fluorinated allylic trichloroacetimidate substrates computed at the PBE1PBE/6-311+G(d,p)-SDD(Rh)-CPCM (toluene) level of theory using the Grimme D3 approximation (empirical dispersion = GD3). Gray, white, red, green, teal, blue, and pale blue ellipsoids represent C, H, O, Cl, Rh, N, and F atoms, respectively. Energies of the methyl-substituted trichloroacetimidate have been reported previously.²⁸

Next, we computed and compared the energies involved in the ionization of the fluoromethyl substituents (Figure 1). The calculated transition energy for the ionization of (S)-fluoromethyl trichloroacetimidate (S)-TS-1-F was 20.0 kcal/mol, showing a considerable change of ~6 kcal/mol when compared to that of its (S)-methyl derivative (S)-TS-1. In addition, the (R)-enantiomer of fluoromethyl trichloroacetimidate (R)-TS4-F is ~1 kcal/mol higher in energy than its methyl analogue, (R)-TS4. Computational calculations indicate that replacing the methyl substituent with a fluoromethyl group increases the energy required for the C–O bond-breaking step, ultimately forming the reactive π-allyl intermediate. This DFT result is consistent with the reactivity pattern of the fluoromethyl-substituted allylic trichloroacetimidate substrate and complements both computational and experimental findings related to the ionization of a trichloroacetimidate bearing an electron-deficient group.⁴⁴ On the basis of these results, one can infer that the observed energy difference best accounts for the sluggish reactivity of the fluorinated trichloroacetimidate substrates. Finally, to account for the observed slight enantioenrichment of the starting material, we hypothesized there may be a differential rate of ionization for (R)- and (S)-fluoromethyl allylic trichloroacetimidate 1. Computational calculations revealed an energy difference of ~3.4 kcal/mol between (R)-TS4-F and (S)-TS-1-F (Figure 1). In contrast, this energy difference was only ~2.0 kcal/mol for methyl-substituted trichloroacetimidate 1a [(R)-TS4 and the (S)-TS-1].

In summary, our approach enables the asymmetric synthesis of challenging tertiary β-fluoroamines using a wide range of anilines and cyclic aliphatic amines. DFT calculations were employed to elucidate the observed difference in reactivity between the fluoromethyl-substituted trichloroacetimidates and their methyl-substituted counterparts. The computational analysis revealed that the ionization transition energy required to generate the reactive π-allyl was higher with the fluorinated trichloroacetimidate substrate. Furthermore, this developed protocol provides efficient direct asymmetric access to tertiary β-fluoroamines in synthetically useful yields and selectivities.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.