Aziridine Opening via a Phenonium Ion Enables Synthesis of Complex Phenethylamine Derivatives

Abstract

We report that the treatment of unsymmetrical 2,3-disubstituted aziridines with TiCl₄ yields β-phenethylamine products via the intermediacy of a phenonium ion. Derivatization of the products obtained via this method is demonstrated. Computational analysis of the reaction pathway provides insight into the reaction mechanism, including the selectivity of the phenonium opening.

Graphical Abstract

The β-phenethylamine motif represents a privileged scaffold in biology.¹ Molecules containing this substructure are important for the treatment of pain, neurological disorders, and many other areas of medicine (Figure 1A). New synthetic methods that enable access to complex β-phenethylamine derivatives are therefore important to develop.² Furthermore, methods that provide β-phenethylamine scaffolds containing additional functional handles for further diversification may accelerate investigations of new chemical space and lead to advances in medicine.³

Figure 1.

Introduction. (A) β-Phenethylamines in biology. (B) Phenonium ions from epoxides. (C) Phenonium-ion-mediated approach to β-phenethylamine derivatives.

As part of a program aimed at exploring the phenonium ion as a key intermediate in synthesis, we recently reported the regiodivergent opening of unsymmetrical phenonium ions derived from benzyl-substituted epoxides (Figure 1B).⁴ Here it was discovered that reaction of epoxide 1 with SnCl₄ or TiCl₄ afforded chiral building blocks 3 and 4, respectively. These reactions were proposed to proceed through the common phenonium ion intermediate 2, the regioselective ring opening of which was under reagent control. With a view toward expanding the utility of phenonium ions in synthesis,^5–7 we questioned whether a related approach could be developed for the synthesis of complex phenethylamine derivatives 6 from simple aziridines 5 (Figure 1C).⁸ The nucleophilic ring opening of aziridines with electron-rich (hetero)aromatics has been demonstrated in inter-⁹ and intramolecular settings;¹⁰ however, no examples of phenonium ion formation from an aziridine have been reported to date.¹¹ Herein we report the development of the phenonium-ion-mediated ring opening of unsymmetrical aziridines for the synthesis of complex phenethylamines. A particular feature of this method is the incorporation of an alkyl halide functional handle into the phenethylamine products.

A key difference between epoxides and aziridines is the presence of the N-substituent, the identity of which dramatically alters the reactivity.¹² Initially, p-methoxybenzyl-substituted aziridine 5a was chosen as a model substrate to evaluate the impact of different N-substituents on the efficiency of the phenonium ion formation and the selectivity of its opening (Figure 2). Treatment of N-Boc aziridine 5a-Boc with TiCl₄ at −78 °C resulted in a 40% yield of the desired phenethylamine 6a-Boc along with a 33% yield of carbamate A (entry 1). N-Ts aziridine 5a-Ts afforded phenethylamine 6a-Ts in 63% yield alongside amine B in 24% yield (entry 2). Employing a more electron-deficient sulfonyl group yielded phenethylamine 6a-Ns in 59% yield and amine B in 10% yield (entry 3). Alkyl-substituted aziridines did not undergo any ring opening, presumably due to their low electrophilicity (entry 4). N-Acyl aziridines resulted in moderate yields of the corresponding phenethylamines (entries 5 and 6). In this stage, N-tosyl aziridine 5a-Ts was chosen to evaluate other reaction parameters due to the ease of substrate synthesis and handling. Running the reaction at 0 °C slightly increased the yield of phenethylamine 6a-Ts (69%, entry 7); however, further temperature increases did not improve the yield (entry 8).

Figure 2.

Reaction optimization. ^aIn situ yield determined by ¹H NMR using 1,3,5-trimethoxybenzene as an internal standard. ^bYield of isolated material. ^c1.1 equiv of TiCl₄ employed.

Having initially optimized the reaction for aziridine 5a-Ts, we wanted to test the reaction with a less electron-rich benzyl substituent. At −78 °C, aziridine 5b-Ts afforded amine C as the sole product (68%, entry 9). Running the reaction at 0 °C provided the desired phenethylamine 6b-Ts in modest yield (23%) with amine C still the major product (61%, entry 10). At room temperature, phenethylamine 6b-Ts became the major product in 43% yield alongside amine C in 34% yield (entry 11). Although the temperature did not have a significant impact on the yield of phenethylamine 6a-Ts, it was clear that it had a greater influence on the reaction of aziridine 5b-Ts. Upon exhaustive re-evaluation of the temperature and N-substituent, it was found that use of N-acyl aziridines at higher temperatures led to a considerable improvement in the reaction efficiency compared with N-tosyl aziridine at the same temperatures (entries 12–15). The best yield was obtained with aziridine 5b-Ac at room temperature (52%, entry 13). Under these conditions, aziridine 5a-Ac afforded the corresponding phenethylamine 6a-Ac in an improved 85% yield of isolated product (entry 14).

With the improved conditions in hand (N-Ac, 0.5 equiv of TiCl₄, CH₂Cl₂, rt or 0 °C), we evaluated the scope of the reaction with respect to the benzyl and alkyl substituents (Figure 3A). The temperature of the reaction (rt or 0 °C) was optimized for each substrate. Rearrangement of aziridines 5 to phenethylamines 6 was achieved with a variety of substituted arenes. For example, 4-methylphenyl (6c), 3,4-dimethylphenyl (6d), and 3,4-dimethoxyphenyl (6e) provided good yields. A 40% yield was obtained when 3-bromo-4-methoxyphenyl was the arene employed (6f). A 4-morpholine-substitued arene afforded the desired phenethylamine 6g in good yield (58%). Indole substrate 5h afforded the corresponding product 6h in 56% yield, and dihydrobenzofuran 5i provided the desired phenethylamine 6i in 60% yield. An ortho-methoxy substituent was also tolerated, giving phenethylamine 6j in 66% yield. The use of highly electron-withdrawing substituents (e.g., trifluoromethyl) did not lead to any phenonium ion formation. This may reflect the lower electrophilicity of aziridines compared with epoxides.¹²

Figure 3.

(A) Substrate scope of the aziridine opening reaction. (B) Demonstration of product utility. ^aYield of isolated material. ^bReaction temperature = 0 °C.

The functional group tolerance of the reaction was also examined. Alkyl-substituted phenethylamines 6k and 6l bearing “amino acid-like” side chains could be accessed from aziridines 5k and 5l in 79 and 42% yield, respectively. Functional groups such as an alkyl chloride (6m), nitrile (6n), and benzyl ether (6o) worked well. A phthalimide-protected amine afforded phenethylamine 6p in modest yield (37%). Lastly, a medicinally relevant piperidine heterocycle was tolerated in the reaction (6q, 53%).

Several derivatization reactions were explored to highlight the potential utility of the phenethylamines obtained using this method (Figure 3B).¹³ Treatment of amine 6a-Ts with KOt-Bu afforded azetidine 7 in 92% yield. Nuclear Overhauser effect experiments confirmed the relative anti stereochemistry between the aryl and alkyl substituents, which is consistent with the inversion of stereochemistry at the aziridine carbon during phenonium ion formation. (See the Supporting Information.) Ruthenium-catalyzed oxidation of the arene to a carboxylic acid affords azetidine 3-carboxylic acid derivative 8. The use of N-acyl aziridines in this method enables ready access to dihydroisoquinoline scaffolds through a Bischler–Napieralski reaction.¹⁴ For example, treatment of amine 6e with POCl₃ afforded cyclized product 9 in 62% yield. Alternatively, the Pictet–Spengler reaction of amine 6a-Ts with paraformaldehyde gave tetrahydroisoquinoline 10 in 56% yield.¹⁵ Notably, in both cyclization reactions, the alkyl chloride remains available for potential further derivatization.

Finally, we wanted to probe the reaction mechanism to gain insight into the factors driving selectivity for product formation. Conventional empirical studies proved impractical due to the short reaction times; therefore, we opted to investigate the transformation computationally by performing a density functional theory (DFT) analysis (Figure 4). A truncated version of parent substrate Int-1 was chosen as the model substrate for this study. We found that coordination of the azridine to TiCl₄ showed an energetic preference for monodentate O-ligation to the acyl group (TiO–Int1) compared with bidentate N,O-ligation to form a titanacycle with an octahedral geometry at the Ti center (TiE–Int1 and TiZ–Int1). Surmountable energy barriers between TiO–Int1 and TiE/Z–Int1 were found (G_rel < 4.0 kcal mol⁻¹), suggesting that these species can readily interconvert (Figure 4A). Although bidentate complexes TiE–Int1 and TiZ–Int1 were slightly higher in energy relative to TiO–Int1, the ΔG^‡_(TS-1) values leading to their respective phenonium ions Int2 were substantially lower: ΔG^‡_(TS-1) = 9.9 kcal·mol⁻¹ from TiE–Int1 and 12.6 kcal·mol⁻¹ from TiZ–TS1 compared with ΔG^‡_(TS-1) = 23.0·mol⁻¹ from TiO–Int1. We next considered chloride opening of the phenonium ion from TiE–Int1 and TiZ–Int1 (Figure 4B). Intramolecular chloride delivery from TiE–Int2 was favored for addition at the less substituted carbon of the spirocyclopropane, leading to Int3–major ΔG^‡ = 13.7 kcal·mol⁻¹) over Int3–minor ΔG^‡ = 16.1 kcal·mol⁻¹), which is consistent with the observed reaction outcomes. From TiZ–Int2, a transition-state structure leading to the major product Int3–major was found ΔG^‡ = 11.8 kcal·mol⁻¹); however, a transition state leading to the minor product could not be located due to geometric constraints of the corresponding phenonium ion. (See the Supporting Information for further discussion.)

Figure 4.

(A) Computed equilibrium between monodentate TiO–Int1 and bidentate TiE– and TiZ–Int1 aziridines and their respective pathways for phenonium ion formation. (B) Computed phenonium ion opening of TiE–Int2. Calculations were performed at the [IEF-PCM(CH₂Cl₂,SMD)-MN15/def2-QZVPP, 0 °C, 1.0 M//IEF-PCM(CH₂Cl₂,SMD)MN15/6-31+G(d),LanL2DZ (for Ti)] level of theory. All G_rel values are referenced to TiO–Int1.

In summary, we have developed a stereospecific transformation of benzyl-substituted aziridines into complex phenethylamine derivatives via the intermediacy of a phenonium ion. The reactions are operationally simple and proceed with fast reaction times (<10 min). DFT analysis of the reaction bolsters the viability of a phenonium ion intermediate and offers insight into the regioselectivity of ring opening. The reaction displays good functional group tolerance, and product utility has also been demonstrated. We anticipate that this reaction will be useful for medicinal chemists looking to explore new chemical space involving phenethylamine motifs. More broadly, this work highlights the synthetic power of phenonium ions as strategic reactive intermediates in the transformation of simple starting materials into complex products.