Enantioselective Synthesis of N-Alkylamines through β-Amino C–H Functionalization Promoted by Cooperative Actions of B(C₆F₅)₃ and a Chiral Lewis Acid Co-Catalyst

Abstract

We disclose a catalytic method for β-C(sp³)–H functionalization of N-alkylamines for synthesis of enantiomerically enriched β-substituted amines, entities prevalent in pharmaceutical compounds and used to generate different families of chiral catalysts. We demonstrate that a catalyst system comprising of seemingly competitive Lewis acids, B(C₆F₅)₃ and a chiral Mg- or Sc-based complex, promotes the highly enantioselective union of N-alkylamines and α,β-unsaturated compounds. An array of δ-amino carbonyl compounds was synthesized under redox-neutral conditions by enantioselective reaction of a N-alkylamine-derived enamine and an electrophile activated by the chiral Lewis acid co-catalyst. The utility of the approach is highlighted by late-stage β-C–H functionalization of bioactive amines. Investigations in regard to the mechanistic nuances of the catalytic processes are described.

Graphical Abstract

1. INTRODUCTION

Enantioselective synthesis of N-alkylamines through the transformations of amino C(sp³)–H bonds has emerged as a powerful strategy to access key building blocks of N-based natural products, drugs and catalysts for stereoselective synthesis.^1–18 A plethora of organometallic catalysts have been introduced that can promote reaction at an α-amino C–H bond by conversion of an amine substrate to appropriately reactive intermediate (e.g., α-amino radical, iminium ion),^19–31 metal–carbenoid insertion,^32–33 or heteroatom-directed metalation.^34–36 More remote γ-amino C–H bonds may be activated by L_nPd-catalyzed and N-directed cyclometalation.^37–41 However, development of methods for synthesis of enantiomerically enriched amines by β-amino C–H functionalization is a major challenge that remains to be addressed.^42–50 α-Amino C–H activation by hydrogen atom or hydride transfer is facilitated by stabilization of the resulting species through hyperconjugation with the nitrogen lone pair. Nonetheless, such processes do not readily occur at the β position of amines. Furthermore, L_nPd-catalyzed β-amino C–H activation requires the presence of a strained 4-membered palladacycle (vs a more favorable 5-membered metallacycle formed by γ-amino C–H activation).^42–44

There are only a limited number of protocols for preparing enantiomerically enriched β-substituted amines through activation of a β-amino C–H bond.^51–54 A notable strategy is Pd/phosphoric acid-catalyzed and IOAc-mediated desymmetrization of gem-dimethyl groups in tetramethyl-morpholinone 1a to give organometallic intermediate I en route to aziridine 2a (Figure 1a).⁵² Equally noteworthy is Pd/PR₃-catalyzed stereospecific cross-coupling of 1,3-oxazinane 1c (prepared by s-BuLi/(+)-sparteine-mediated α-lithiation of 1b and Li/Zn exchange with ZnCl₂) and Ph–Br to give 2c via L_nPd–enamine II (Figure 1b).⁵⁴ Still, key shortcomings remain unaddressed. For instance, acid–base complexation often occurs in a mixture that contains a Lewis acidic catalyst (e.g., Pd(OAc)₂) and a Lewis basic N-alkylamine substrate and/or product which may also contain other Lewis acid-sensitive functional groups.^55–62 Additionally, because these methods require IOAc or s-BuLi, moieties that readily react with oxidants and Lewis bases can be problematic. As a result, substrate scope is confined to highly sterically encumbered and/or Boc-protected amines (1a, 1b and analogues) with minimal electronic and steric affinity towards Lewis acids, Lewis bases or oxidants. Development of a sustainable element-based, highly functional group tolerant and enantioselective catalyst system which allows access to β-substituted N-alkylamines via β-amino C–H functionalization under redox-neutral conditions therefore constitutes a compelling research objective.

Figure 1.

Enantioselective Transformations of β-Amino C–H Bonds

Conversion of β-amino C–H bonds can be achieved nonstereoselectively through in situ generation of enamines from N-alkylamines;^63–85 but enamine precursors are largely limited to N,N-dialkylanilines. In the case of more Lewis basic trialkylamines and substrates containing acid- and/or base-sensitive moieties, catalyst deactivation can become problematic.^55–62 One strategy to overcome mutual quenching would be to use strongly Lewis acidic and sterically hindered B(C₆F₅)₃^86–88 to convert trialkylamines into enamines;^63–76 this would involve (F₅C₆)₃B-mediated hydride abstraction from the amine to give a borohydride/iminium complex,^63–76 which would in turn be deprotonated to yield an enamine (Figure 1c, 1 → III → IV).^89–92 Chang and co-workers have shown that B(C₆F₅)₃ is capable of promoting β-silylation of N-arylpiperidines to generate rac-bridged sila-N-heterocycles.⁶⁴ We have introduced a method for (F₅C₆)₃B-catalyzed hydrogen isotope exchange involving β-amino C–H bonds of various bioactive molecules.⁶⁵ Still, engagement of the enamine intermediates formed by (F₅C₆)₃B/Brønsted base-catalyzed dehydrogenation of N-alkylamines has not been utilized in an enantioselective transformation.

In contemplating ways to develop a protocol that involves β-C–H alkylation of N-alkylamines (1) through an enantioselective union with α,β-unsaturated compounds (3), we envisioned using a set of B(C₆F₅)₃, a Brønsted base catalyst, and a chiral Lewis acid co-catalyst (M–L*) that might be induced to function cooperatively (Figure 1c).^{63–76, 93–102} We imagined that B(C₆F₅)₃ being the recipient of a hydride from an amine (1), leading to the formation of a borohydride and an iminium ion (III). A Brønsted base catalyst would subsequently deprotonate III to furnish enamine IV. An ensuing enantio- and diastereo-selective C–C bond formation between enamine IV and α,β-unsaturated compound, activated by the chiral Lewis acid co-catalyst (V), would deliver zwitterionic VI. This would be followed by protonation and reduction to give β-alkylation product 4. A key advantage of using untethered and independently operational catalysts is that efficiency and stereoselectivity may be easily optimized by evaluation of readily accessible Lewis acids, Lewis bases and chiral ligands (vs bifunctional catalysts that require tethering of different catalytic sites). However, B(C₆F₅)₃ and the chiral Lewis acid co-catalyst must be able to perform their independent roles without overlapping functions, as otherwise, B(C₆F₅)₃ could promote the racemic reaction through activation of both substrates, most likely resulting in diminished enantioselectivity. Here, we report that functionally similar B(C₆F₅)₃ and a chiral Lewis acid co-catalyst can operate in concert to engender enantioselective coupling of N-alkylamines and α,β-unsaturated compounds.

2. RESULTS AND DISCUSSION

2.1. Method Development

2.1.1. Identification of Optimal Conditions.

To begin, we set out to investigate if B(C₆F₅)₃ might activate both N,N-dibenzylethanamine (1d) and diisopropyl fumarate (3d), generating 4d (Table 1). Treatment of 1d and 3d with 10 mol % B(C₆F₅)₃ and 10 mol % Et₃N, 2,2,6,6-tetramethylpiperidine (TMP) or 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) afforded 4d in 50%, 25%, and <5% yield, respectively (entries 1–3). Other than 4d, over alkylation product 5d was also formed (entries 1–6), probably through in situ conjugate addition of an enolate (VI, Figure 1c) to 3d. When the transformation was carried out without a Brønsted base, 4d (73% yield) and 5d (25% yield) were formed more efficiently (entry 4), suggesting that 1d, 4d and/or 5d can deprotonate an iminium to form an enamine (III → IV, Figure 1c). To suppress formation of 5d, the mixture was diluted (entries 5–6); by using 10 mol % of B(C₆F₅)₃ and 0.80 mL of benzene (vs 0.20 mL used in entries 1–4), we were able to obtain 4d (91% yield) within 12 hours (entry 6).¹⁰³ Moreover, no 4d was formed when less hindered BCl₃ or less acidic BPh₃ were involved (entries 7–8). Without B(C₆F₅)₃, 4d was not generated (entry 9). These findings support the notion that strongly Lewis acidic B(C₆F₅)₃ in combination with sterically demanding and electron-rich N-alkylamines constitute the most effective catalyst–substrate combination.

Table 1.

Evaluation of Various Reaction Parameters^a,b,c

entry	Lewis acid	Brønsted base	time (h)	yield of 4d (%)	yield of 5d (%)
1	B(C6F5)3	Et3N	3	50	23
2	B(C6F5)3	TMP	3	25	29
3	B(C6F5)3	DBU	3	<5	<5
4	B(C6F5)3	none	3	73	25
5d	B(C6F5)3	none	3	54	5
6d	B(C6F5)3	none	12	91	7
7	BCI3	none	12	0	0
8	BPh3	none	12	0	0
9	none	none	12	0	0

^aConditions: N,N-dibenzylethanamine (1d, 0.20 mmol), diisopropyl fumarate (3d, 0.30 mmol), Lewis acid, Brønsted base, C₆H₆ (0.20 mL), under N₂, 50 °C.

^bYield was determined by the ¹H NMR analysis of unpurified reaction mixtures with m-xylene as the internal standard.

^cThe structure and relative configuration of 5d were established by the nuclear Overhauser effect spectroscopy (NOESY) studies.

^d0.80 mL of C₆H₆ was used.

2.1.2. Scope.

Many acyclic and cyclic N-alkylamines may be used in the reaction with diisopropyl fumarate 3d to generate the corresponding β-substituted amines (4d–4o, Figure 2). Reaction with N,N-dibenzylethanamine 1d and 3d afforded 4d in 88% yield. N-Benzhydryl-substituted secondary amines 1e and 1f were suitable substrates, furnishing 4e (93% yield) and 4f (87% yield, 2.1:1 dr), respectively. None of the 1,4-addition product was observed in the case of a secondary amine bearing a more hindered trityl group (e.g., N-tritylethanamine); this may be attributed to rapid substrate decomposition. With less hindered N-benzylethanamine, the formation of (F₅C₆)₃B–amine adduct may compete with Lewis acid-catalyzed hydride abstraction, resulting in minimal formation of the desired product. We then investigated reactions with different chiral N-alkylamines (1g–1i, Figure 2). Accordingly, 4g was produced in 90% yield as a 1.4:1 mixture of easily separable diastereomers, allowing us to secure the amino ester in enantiomerically pure form. As indicated by synthesis of 4h and 4i, chiral pyrrolidines may be used as starting materials. Whereas the reaction with the less hindered 1-(4-methoxy-2,6-dimethylphenyl)piperidine furnished 4j as the only product in 89% yield, with the bulkier 4,4-dimethyl-substituted piperidine substrate, enamine 4k was obtained in 93% yield.¹⁰⁴

Figure 2. β-Alkylation of Different N-Alkylamines.

The values correspond to yields of isolated and purified products. ^a Conditions: N-alkylamine (1, 0.20 mmol), diisopropyl fumarate (3d, 0.30–0.40 mmol), B(C₆F₅)₃ (10 mol %), C₆H₆, under N₂, 50 °C. ^b Yield of isolated and purified product. The dr values were determined by the ¹H NMR analysis of the unpurified reaction mixtures. ^c The structure and relative configuration of 4f, 4j, 4l, and 4m were assigned in analogy. The stereochemistry of 4n and 4o was assigned in analogy and also by the NOESY studies. ^d The absolute configuration of 4g was determined based on the specific rotation of the derivative. ^e The structure and absolute configuration of 4i was established by the X-ray crystallographic analysis. The stereochemistry of 4h was assigned in analogy and also by the NOESY studies. See the Supporting Information for details.

The method is applicable to late-stage modification of N-containing bioactive molecules that possess an array of Lewis acid-sensitive functional groups (1l–1n; Figure 2). In addition to the N-alkylamine moieties of 1l–1n, a ketone (1l), a benzothiophene (1l), an ether (1l, 1n), a pyrimidinone (1m) and a benzoisoxazole (1m) were tolerated. Thus, structures of piperidine-based compounds, such as raloxifene (osteoporosis treatment), risperidone (anti-psychotic) and paroxetine (antidepressant) could be readily altered. Silyl-protected raloxifene 1l reacted efficiently with 3d to afford 4l in 80% yield and 2.5:1 dr. Risperidone 1m, possessing more sterically hindered β-amino C–H bonds (vs. 1l), was merged with 3d, furnishing enamine 4m (32% yield); minimal amounts (<5%) of the product derived from alkylation of acyclic β-amino C–H bond (H labelled in red) could be observed (¹H NMR analysis). The process involving N-benzyl paroxetine 1n and 3d afforded 4n and 4o in 39% and 14% yield, respectively.

2.1.3. Diastereo- and Enantio-selective Processes.

To develop a highly diastereoselective and catalytic β-amino C–H alkylation variant, we chose to utilize the prolinol derivative 1i^105–108 and (S,E)-4-phenyl-3-(4,4,4-trifluorobut-2-enoyl)oxazolidin-2-one 6a as model substrates (Figure 3a). In the event, (F₅C₆)₃B-catalyzed reaction of 1i and diisopropyl fumarate 3d gave β-alkylation product 4i in only 56% yield and 2.5:1 dr (Figure 2). To probe whether the use of enantiomerically pure electrophile leads to improved dr, we reacted 1i and 6a in the presence of B(C₆F₅)₃ to find that product 7a was formed in <5% yield. These findings suggest that B(C₆F₅)₃ facilitates the conversion of 1i into an enamine but does not sufficiently activate 6a. We therefore decided to evaluate a blend of B(C₆F₅)₃ and various Lewis acid co-catalysts, latter of which may activate 6a, to establish that, with 10 mol % B(C₆F₅)₃ and ZnI₂, 7a-(R,R,R,S) can be obtained in 82% yield and as a single diastereomer (>20:1 dr; see the Supporting Information for details). The presence of HBPin led to enhanced dr, as 7a was isolated in 67% yield and 10:1 dr in its absence (see the Supporting Information for details). We also explored the suitability of developing an enantioselective β-amino C–H alkylation reaction between achiral substrates 1o and 6b, promoted by a combination of B(C₆F₅)₃ and an enantiomerically pure organometallic co-catalyst (Figure 3b). We discovered that, with a complex of PyBOX L1–Mg(ClO₄)₂, β-amino C–H alkylation of 1o proceeds to afford 7b in 67% yield (1.6:1 dr, up to 90:10 er). However, neither efficiency nor enantioselectivity could be improved by further catalyst optimization (with 6b as electrophile; see the Supporting Information for details).

Figure 3. Stereoselective β-Functionalizations of N-Alkylamines.

The values correspond to yields of isolated and purified products. See the Supporting Information for details.

To improve enantioselectivity, we probed the transformations involving a number of chiral Lewis acid co-catalysts and α,β-unsaturated compounds bearing different auxiliaries.^109–112 We found that 1-(4-methoxyphenyl)pyrrolidine 1p reacts efficiently with 2-acryloylpyrazolidinone derivative 6c in the presence of 10 mol % of B(C₆F₅)₃ together with Sc(OTf)₃ and various chiral bisoxazoline ligands (Figure 4). We then found that reactions with Ph–BOX, Ph–DBFOX and Ph–PyBOX ligands (e.g., L2–L4) are hardly enantioselective (58:42–50:50 er). The situation improved considerably with alkyl-substituted PyBOX ligands (e.g., L5–L8), and in the presence of (S)-i-Bu–PyBOX (L8), 7c was formed in 72% yield, 2.7:1 dr and up to 98:2 er. Reaction efficiency improved (85% yield) when L9 and its diastereomer L10 were used. The stereochemical course of 1p-derived enamine addition to [L10–Sc(OTf)₃]-activated 6c can be rationalized by the models presented in Figure 5.^113–115 Models VII and VIII represent the energetically minimized structures of [L10–Sc(OTf)₃] docked with 6c.^113–115 As shown in model VII, the high level of enantioselectivity observed in the formation of 7c-(R,S) can be explained by selective 1,4-addition of the enamine to the re-face of [L10–Sc(OTf)₃]-bound 6c; as depicted in model VIII, the si-face is effectively shielded by a sec-butyl group of the PyBOX ligand.

Figure 4. Evaluation of Chiral Ligands.

Yield and diastreomeric ratio (dr) values were determined by the ¹H NMR analysis of unpurified reaction mixtures with mesitylene as the internal standard. Enantiomeric ratio (er) values were determined by the HPLC analysis of isolated and purified product. Conditions: N-arylpyrrolidine (1p, 0.20 mmol), α,β-unsaturated compound (6c, 0.30 mmol), B(C₆F₅)₃ (10 mol %), Sc(OTf)₃ (5.0 mol %), ligand (6.0 mol %), CH₂Cl₂ (2.0 mL), under N₂, 60 °C, 1 h. See the Supporting Information for details.

Figure 5.

Stereochemical Rationale for Enamine Addition to α,β-Unsaturated Compounds Catalyzed by [L10–Sc(OTf)₃].

Enantioselective reactions with an array of N-alkylamines were carried out in the presence of B(C₆F₅)₃, a L–Sc(OTf)₃ complex and 6c–6f (Figure 6). β-Alkyl derivatives of 1-(4-methoxyphenyl)pyrrolidine (1p) bearing γ-ester, ketone or CF₃ groups (7c–7f) were thus synthesized in 62–83% yield, 2.0:1–6.7:1 dr and 95:5–98:2 er. The reaction of 1p with 6c (R⁵ = Et) and 6d (R⁵ = Bn) gave 7c (2.0:1 dr, up to 98:2 er) and 7d (3.0:1, up to 95:5 er), respectively, indicating that N-substituents of the pyrazolidinone unit have influence over both diastereo- and enantio-selectivity. When acyclic N-ethyl-4-methoxy-N,2,6-trimethylaniline 1q was reacted with F₃C-substituted 6f (Figure 7a), N-arylpiperidine derivative 7g was produced in 67% yield (>20:1 dr, 95:5 er); generation of 7g probably entails β-amino C–H alkylation of 1q by 6f to afford a zwitterionic intermediate containing an iminium and an enolate (IX), followed by isomerization of the iminium to give X; ensuing intramolecular Mannich-type reaction gives 7g. The unions of N-arylpiperidines (1j–1k) or N-arylazapene (1r) with 6f afforded enamines 7h–7j in 74–95% yield and 94:6–96:4 er. The reaction of N-arylpiperidine-3,3,5,5-d₄ 1j-d (0.10 mmol) and 6f (0.20 mmol) gave 7h-d (84% yield, 95:5 er) and 8h-d (0.08 mmol; Figure 7b). Spectroscopic analysis of 7h-d and 8h-d revealed that there is deuterium incorporation at their enolizable α-carbonyl units, and that there is D/H exchange at C5 of 7h-d (>95% in 1j-d → 81% in 7h-d).⁶⁵ These results imply that in situ generated [(F₅C₆)₃B–H]⁻ [base–D]⁺ reacts with 6f to produce 8h-d to regenerate B(C₆F₅)₃ (vs by releasing H–D). We were able to functionalize the β-Amino C–H bonds of bioactive trialkylamines, including cloperastine (cough suppressant) 1s and raloxifene 1l, to generate enamines 7k and 7l in 53% yield (90:10 er) and 57% yield (95:5 er), respectively.

Figure 6. Stereoselective Processes.

Cooperative functions of B(C₆F₅)₃ and L–Sc(OTf)₃ promote stereoselective β-alkylation of N-alkylamines. ^a Conditions: N-alkylamine (1, 0.10 mmol), α,β-unsaturated compound (6, 0.15–0.20 mmol), B(C₆F₅)₃ (10 mol %), L10–Sc(OTf)₃ (10 mol %), CH₂Cl₂ (1.0 mL), under N₂, 60 °C, 1–36 h. ^b 10 mol % of L6–Sc(OTf)₃ was used. ^c 20 mol % of B(C₆F₅)₃ was used. ^d 20 mol % of L10–Sc(OTf)₃ was used. ^e The reaction was performed at 80 °C. ^f 10 mol % of L8–Sc(OTf)₃ was used. See the Supporting Information for details.

Figure 7. Additional Data Interpretation.

(a) Sequential β-alkylation and Mannich-type reaction gave piperidine derivative 7g. (b) D⁺ derived from 1j-d was incorporated into enolizable positions of 7h-d and 8h-d.

2.1.4. Scalability and Modifications of β-Functionalized Amines.

The catalytic method is scalable. For example, treatment of 4.0 mmol of N-arylpiperidine 1j and 6f with 10 mol % B(C₆F₅)₃ and 10 mol % L10–Sc(OTf)₃, (CH₂Cl₂, 12 h, 60 °C) afforded 7h in 95% yield (1.9 g; Figure 8a). Treatment of 7f-(R,S) with LiSEt followed by reduction of the resulting thioester with LiAlH₄ furnished alcohol 9a-(R,S) in 95% overall yield (Figure 8b). Compound 9a-(R,S) was subsequently converted to its derived carbamate 9b-(R,S) which was subjected to the X-ray crystallographic analysis for determination of absolute configuration (see the Supporting Information for details). Enamines obtained by enantioselective β-alkylation were found to be versatile intermediates (Figure 8c). Hydrogenation of enamine 7h by a chiral Ir-based catalyst afforded 9c in 83% yield and 5.0:1 dr,¹¹⁶ and treatment of 7h with NFSI and TMSCN gave fluorocyanation product 9d in 88% yield and 4.0:1 dr.¹¹⁷

Figure 8. Modification of Products and Scalability.

(a) The enantioselective reaction is amenable to gram-scale operations. (b) Transformation of pyrazolidinone auxiliary into thioester, alcohol and then to carbamate was achieved. Absolute configuration of 9b was obtained through X-ray crystallographic analysis. (c) Versatility of enamine 7h was demonstrated through catalytic hydrogenation and fluorocyanation reactions. See the Supporting Information for details.

2.2. Mechanistic Investigations

We designed and performed studies to gain insight regarding the mechanistic nuances of the catalytic process.¹¹⁸ These studies included determining reaction orders, kinetic isotope effects, and Hammett ρ values (Figures 9, 12, and 13, respectively). Additionally, these investigations led to revised pathways for catalytic β-C–H alkylation reaction (Figures 10–11).

Figure 9. Determination of Reaction Orders.

Log(rate) vs log(concentration) plot is employed to determine the reaction order for: (a) B(C₆F₅)₃. (b) L6–Sc(OTf)₃. (c) amine 1p. (d) α,β-unsaturated compound 6g. See the Supporting Information for details.

Figure 12. Kinetic Isotope Effect Studies.

The independent rate measurement studies indicate that intramolecular proton transfer is the turnover-limiting step. See the Supporting Information for details.

Figure 13. Hammett Studies.

Hammett plots of rates for the reaction of N-aryl substituted pyrrolidines 1 and aryl substituted α,β-unsaturated compounds 6. (a), (c) N-Arylpyrrolidine derivatives 1 that contain electron-donating substituents reacted more rapidly. (b), (d) α,β-Unsaturated compounds 6 that contain electron-withdrawing substituents reacted more rapidly. See the Supporting Information for details.

Figure 10. A Catalytic Cycle Consistent with the Results of Mechanistic Investigations.

Kinetic and NMR studies indicate that the turnover-limiting step is intramolecular proton transfer step.

Figure 11. Origin of Enamine and N-Alkylamine Products.

(a), (b) Reaction progress analysis of the coupling of 1j and 3d. (c) 5j and/or 1j-d can serve as H⁺ or D⁺ and hydride source in (F₅C₆)₃B-catalyzed transfer hydrogenation of 5j. See the Supporting Information for details.

2.2.1. Determination of Reaction Orders and a Consistent Mechanistic Pathway.

We found that the reaction of N-arylpyrrolidine 1p with α,β-unsaturated compound 6g has a second-order dependence on B(C₆F₅)₃ concentration (Figure 9a) but is not at all impacted by the concentration of L6–Sc(OTf)₃ complex (Figure 9b). Furthermore, we found a first-order dependence on amine concentration of 1p (Figure 9c) but, a reverse first-order dependence on the concentration of electrophile 6g (Figure 9d). The independence of the reaction rate on the initial concentration of L6–Sc(OTf)₃ suggests that enantioselective C–C bond forming event between in situ generated enamine and [L6–Sc(OTf)₃]-activated 6g (Figure 10, XV → XVI) occurs after the turnover-limiting step. Furthermore, the negative first-order dependence on the concentration of 6g implies that the resting state consists of 6g and the Lewis acid. Spectroscopic analysis of the reaction mixture (¹⁹F NMR) supports the proposal in regard to formation of [6g–B(C₆F₅)₃] (XI).¹¹⁹

Thus, it is likely that β-C–H alkylation proceeds by the release of B(C₆F₅)₃ from XI, which then abstracts a hydride from 1 to form an iminium/borohydride complex (XII). Borohydride reduction of (F₅C₆)₃B-activated 6 delivers a [(F₅C₆)₃B–enolate]⁻[iminium]⁺ complex (XII → XIII).^{28, 120} Subsequent irreversible isomerization of the iminium into an enammonium (XIII → XIV) is likely the turnover-limiting step (see the following kinetic isotope effect and Hammett studies, as well as the Supporting Information for details).^121–124 Protonolysis of [(F₅C₆)₃B–enolate]⁻ in XIV releases B(C₆F₅)₃, 8, and an enamine (XV). Then, enantioselective C–C bond forming reaction between the enamine and [L–Sc(OTf)₃]-activated electrophile 6 leads to a zwitterionic intermediate that bears an iminium and an enolate moiety (XV → XVI). Finally, proton transferwithin XVI produces 7-enamine and regenerates L–Sc(OTf)₃, thus closing the cycle. Alternatively, borohydride reduction of the iminium and protonation of the resulting enolate in XVI produces an N-alkylamine product (7-alkylamine), as illustrated by the studies described below.

2.2.2. Origins of Enamine and N-Alkylamine Products.

β-Amino C–H alkylation products (Figures 2 and 6) were obtained either as an enamine, an N-alkylamine, or a mixture of the two (e.g., 4n and 4o, Figure 2). We wondered if the N-alkylamine products were formed by transfer hydrogenation of XVI (XVI → 7-alkylamine, Figure 10), or whether they were generated through reduction of enamines (7-enamine → 7-alkylamine). To identify each product’s origins, we studied the progress of (F₅C₆)₃B-catalyzed reaction between N-arylpiperidine 1j and 3d (Figure 11a) which gives a mixture of N-alkylamine 4j (57% yield) and enamine 5j (29% yield) using 0.4 mL of C₆D₆ (vs the process involving 1.6 mL of C₆H₆ which selectively gives 4j; Figure 2). We found that there is minimal transformation of 5j into 4j as evidenced by the mostly unchanged concentration of 5j once the β-alkylation reaction completed (2 h; Figure 11b). These results are consistent with the scenario that 7-alkylamine (Figure 10a) is formed by transfer hydrogenation of XVI without the intermediacy of 7-enamine. Nonetheless, the source of proton and hydride remained to be identified.

To probe further if, and under precisely what conditions 5j can be converted to 4j, we investigated the transformation of 5j (isolated and purified by flash silica gel chromatography) in the presence of 20 mol% B(C₆F₅)₃ (C₆H₆, 12 h; Figure 11c). This led us to observe that there was 30% conversion to 4j; in addition, [pyridinium]⁺[C₆F₅]⁻ (9j, 15 %) was also produced. Treatment of 5j with N-arylpiperidine-3,3,5,5-d₄ 1j-d furnished 4j-d in 50% yield, and not only was there significant deuterium incorporation at C3 and C5 positions of 4j-d (71% and 67%, respectively), recovered 1j-d had also undergone D/H exchange. These data suggest that B(C₆F₅)₃ can catalyze transfer hydrogenation of 5j in the presence of another molecule of 5j and/or 1j-d serving as sources of H⁺ (or D⁺) and hydride (Figure 11c). Nevertheless, under the standard conditions for (F₅C₆)₃B-catalyzed β-C–H alkylation reaction (Figures 10 and 11a), in situ generated [(F₅C₆)₃B–H]⁻[Base–H]⁺ (derived from the reaction of B(C₆F₅)₃ and N-alkylamine 1) appears to react with either a highly reactive zwitterionic intermediate (XVI → 7-alkylamine) or (F₅C₆)₃B-activated α,β-unsaturated compounds (XII → XIII → XIV → 8). As a consequence, hydrogenation of the relatively unreactive enamine 5j to give 4j may be outcompeted by these more facile processes.

2.2.3. Kinetic Isotope Effect Studies.

To shed light on the hydride abstraction (Figure 10, 1 → XII), and deprotonation steps (XIII → XIV → XV), deuterium-labeled N-arylpyrrolidines 1t-d and 1u-d were prepared, and their reactions with 6g were probed (Figure 12). Based on the aforementioned rate studies (Figure 9), hinting that the turnover-limiting is prior to the stereoselective C–C bond forming process (Figure 10, XV → XVI), the overall rate of the reaction can be affected for reactions involving both α-deuterated 1t-d and β-deuterated 1u-d.^125–126 However, independent rate measurements involving 1p and 1t-d (Figure 12a) were found to have k_H/k_D = 1.28 ± 0.07.¹²⁷ On the other hand, comparison of the reaction rate between 1p and 1u-d (Figure 12b) revealed that 1p reacts 2.5 times faster than 1u-d (k_H/k_D = 2.50 ± 0.13).¹²⁷ These KIE experiments support the notion that the turnover-limiting step is the conversion of iminium into enammonium (Figure 10, XIII → XIV) which entails the cleavage of α-imino C–H or C–D bonds; furthermore, (F₅C₆)₃B-catalyzed hydride abstraction (1 → XII) and the following borohydride reduction (XII → XIII) steps are reversible, thus leading to the equilibrium KIE of 1.28.

2.2.4. Hammett Studies.

Hammett studies revealed a strong dependence of the reaction rate on the electronic properties of the N-alkylamines, with N-arylpyrrolidine derivatives (1) bearing electron-donating substituents reacting more rapidly (ρ = −4.9, Figures 13a and 13c). The large negative ρ value obtained supports the proposed mechanism (Figure 10) in which B(C₆F₅)₃ abstracts a hydride from N-arylpyrrolidine 1 into a N-aryl iminium cation (1 → XII), and its isomerization into an enammonium species (XII → XIII → XIV); these processes take place at or prior to the turnover-limiting step. While the reaction rate was found to be less dependent of the electronic properties of α,β-unsaturated compounds 6, those involving more electron-withdrawing groups reacted with higher efficiency (ρ = 0.92, Figures 13b and 13d). This latter outcome is congruent with the hypothesis that 6 reacts with in situ generated [(F₅C₆)₃B–H]⁻ to afford a boron–enolate intermediate (Figure 10b; XII → XIII), and that this hydride transfer also occurs at or prior to the turnover-limiting step.

3. CONCLUSIONS

In brief, we have developed an efficient catalytic method for functionalization of β-amino C–H bonds to generate enantioenriched δ-amino carbonyl compounds. We find that by using a blend of B(C₆F₅)₃ and a chiral Sc-based complex, it is possible to convert an N-alkylamine into an enamine and then promote its enantio- and diastereo-selective reaction with an α,β-unsaturated compound. The catalyst system is tolerant of a wide variety of Lewis acid-sensitive functional units and therefore applicable to late-stage modification of relatively complex (and bioactive) trialkylamine molecules. Mechanistic investigations reveal that the turnover-limiting step is probably isomerization of N-alkylamine-derived iminium ion into an enammonium intermediate.

The principles outlined above demonstrate that proper combination of an achiral organoborane and an enantiomerically pure organometallic catalyst may be used for chemo- and enantioselective C–H bond functionalization, providing a rational basis for future development of processes for late-stage stereoselective β-functionalization of multifunctional bioactive amines. Studies aimed at achieving these objectives are underway.