Regio- and Enantioselective Synthesis of Trifluoromethyl-Substituted Homoallylic α-Tertiary NH₂-Amines by Reactions Facilitated by a Threonine-Based Boron-Containing Catalyst

Abstract

A method for catalytic regio- and enantioselective synthesis of trifluoromethyl-substituted and aryl-, heteroaryl-, alkenyl-, and alkynyl-substituted homoallylic α-tertiary NH₂-amines is introduced. Easy- to-synthesize and robust N-silyl ketimines are converted to NH-ketimines in situ, which then react with a Z-allyl boronate. Transformations are promoted by a readily accessible L-threonine-derived aminophenol-based boryl catalyst, affording the desired products in up to 91% yield, >98:2 α:γ selectivity, >98:2 Z:E selectivity, and >99:1 enantiomeric ratio. A commercially available aminophenol may be used, and allyl boronates, which may contain an alkyl-, a chloro-, or a bromo-substituted Z-alkene, can either be purchased or prepared by catalytic stereoretentive cross-metathesis. What is more, Z-trisubstituted allyl boronates may be used. Various chemo-, regio-, and diastereoselective transformations of the α-tertiary homoallylic NH₂-amine products highlight the utility of the approach; this includes diastereo- and regioselective epoxide formation/trichloroacetic acid cleavage to generate differentiated diol derivatives.

Introduction

Catalytic enantioselective transformations that furnish α-tertiary amines,^[1,2] fragments found in many bioactive molecules (Scheme 1a), are small in number, rendering their development a compelling research objective. Strategies for preparation of α-tertiary homoallylic NH₂-amines with a trifluoromethyl and an aryl substituent are especially desirable, as the resulting products can be converted to a variety of organofluorine compounds that cannot be easily prepared otherwise in high diastereo- and enantioselectivity. Among the available approaches, there are reactions of an enantiomerically pure N-tert-butylsulfinyl ketimine with an allylzinc or a costly allyl–In compound (Scheme 1b).^{[ 3, 4 ]} Two other strategies are catalytic enantioselective processes involving in situ generated allyl species and N-benzyl substrates (Scheme 1c).^[5,6] Products either bear an alkene with an internal carboxylic ester unit^[5] or an (E)-β-substituent.^[6,7] These are significant advances no doubt, but the following issues remain unaddressed: 1) Direct formation of unprotected α-tertiary homoallylic amines would obviate the need for subsequent release of the NH₂-amine by acid treatment, which can be detrimental to the structural integrity of a sensitive benzylic C–N bond.^[8] 2) Availability of catalysts that can be readily modified/optimized and do not require a costly and/or precious metal. 3) Regio- and diastereoselective product modification, a characteristic more easily accommodated with Z-homoallylic amine (vs. an E- or a monosubstituted olefin).^[9]

Scheme 1.

Bioactive compounds that contain a F₃C- and aryl-substituted a-tertiary amine moiety and methods that may be used to prepare them. Abbreviations = BACE-1: b-secretase-1; MGAT: monoacylglycerol acyltransferase.

A problem with directly accessing NH₂-amines is that F₃C-substituted NH-ketimines are unstable (moisture-sensitive).^[10] We surmised that one might utilize N-protected ketimines that are readily available, robust, and can be converted to NH-ketimines in situ. We thus chose to investigate N-trimethylsilyl ketimines, stable entities accessible in multigram quantities by reactions between ketones and lithium bis(trimethylsilyl)amide (1.5 h, 50–95% yield) ^[11] in sufficiently high purity for N-trimethylsilyl ketimines to be directly usable (no purification). Whether in situ NH-ketimine formation would occur readily or the minimally electrophilic NH-ketimines would react efficiently and enantioselectively, we did not know. As catalysts, we opted for aminophenol-based boryl systems, easily to prepare and modifiable entities that at the time had been used for enantioselective additions to aldimines,^[12] ketones, ^[13] and aldehydes,^[14] but not the far less reactive NH-ketimines.

Results and Discussion

We began by probing the possibility of synthesizing 2a (Scheme 2a) by reaction of silyl ketimine 1a and allyl–B(pin) (pin = pinacolato) with 5.0 mol % ap-1a under the standard conditions (e.g., 5.0 mol % Zn(OMe)₂, 1.5–2.5 equiv. alcohol, tol., 0–22 °C), but none of 2a was formed (<2%). We then examined the influence of different fluorides for in situ NH-ketimine generation,^[15] establishing that with 5.0 mol % tetra-n-butylammonium difluorotriphenylsilicate (tbat, (nBu)₄N(Ph₃SiF₂)), 2a can be generated in 94% yield and 91:9 enantiomeric ratio (e.r.). This was despite the fact there was 35% conversion to rac-2a without an aminophenol. By monitoring the transformation spectroscopically (¹⁹F NMR; see data in Scheme 2a), we found that the NH-ketimine (1a’) is formed rapidly, indicating that silyl removal occurs in the course of the reaction (i.e., the addition does not directly involve silyl-ketimine 2a). Investigation of other aminophenol ligands and/or silyl ketimines, did not result in any significant improvement in e.r.^[15]

Scheme 2.

Initial studies involving allyl–B(pin) and new challenges regarding reaction with NH-ketimines. See the Supporting Information for details. Abbreviation: tbat = (nBu)₄N(Ph₃SiF₂).

We found it intriguing that the e.r. values in Scheme 2a are lower than those for allyl additions to F₃C-substituted ketones (92.5:7.5–96:4 e.r.).^[13b,16] For a ketone (I, Scheme 2b), electrostatic attraction involving the ammonium unit diminishes electron–electron repulsion between the carbonyl oxygen and a C–F bond.^[13a] With a ketimine, on the other hand, reaction via II or III can lead to the major enantiomer, but in the former (II), electrostatic attraction (see I and III) is not feasible because of the NH proton; steric factors in II thus become central and transformation via IV more competitive. In line with this scenario, when Ph₃Si-substituted ap-1b (Scheme 2b) was used, 2a was formed in 72:28 e.r. (89% yield).

The above findings insinuated that it might be possible to exploit steric factors to improve e.r. We recently showed that additions of Z-disubstituted allyl boronates to ketones or aldehydes are more enantioselective probably because the substituent at the stereogenic carbon center of the chiral allyl boronate (e.g., Me unit in VI, Scheme 3a) prefers to be oriented pseudo-axially, away from the catalyst framework.^[13d,14] Accordingly, we investigated the reaction with Z-crotyl– B(pin) ((Z)-3). Control experiments indicated that the non-catalytic pathway, while completely γ-selective, is inefficient (12% conv.; Scheme 3a), and, indeed, with 5.0 mol % ap-1a there was >98% conversion to 4a, which was formed in 96:4 e.r. (vs. 91:9 e.r. for 2a). Consistent with the proposed model, and similar to previous cases,^[13c] reaction with (E)-3 was inefficient (<10% conv. under otherwise identical conditions). Nevertheless, α:γ selectivity was moderate (82:18; 73% yield of the α isomer), implying that conversion of V to lower energy VII by borotropic shift and addition via VIII (Scheme 3a) to generate the γ isomer competes with reaction via VI to give 4a.

Scheme 3.

Catalytic reactions with Z-crotyl–B(pin) are highly α- and enantioselective as a result of several favorable steric and electronic factors originating from O-methyl threonine-derived catalyst. See the Supporting Information for details.

We reasoned that one way to improve the α:γ selectivity might be to enhance the Lewis acidity of a chiral catalyst’s B center. Our hope was that by enhancing ketimine activation, C–C bond formation could be accelerated to a greater extent than the alternative borotropic shift (V→VII). Somewhat surprisingly, however, a smaller (expected to lower steric repulsion in II) and electron-withdrawing CF₃ unit in ap-1c (Scheme 2b)^[13e] adversely impacted the e.r. We therefore chose to investigate a threonine-based catalyst based on the logic that the C–O bond neighboring the C–N bond might increase Lewis acidity of the boron center and the ammonium unit to enhance electrostatic attraction (III, Scheme 2b) and facilitate catalyst-ketimine association (pre-empting borotropic shift).

With O-methyl-L-threonine derived ap-2a (Scheme 3b), 4a was generated in 97:3 α:γ selectivity (vs. 82:18, ap-1a) without any diminution in Z:E ratio or e.r. Equally important, with diastereomeric ap-2b, derived from O-methyl-D-threonine, 4a was formed in lower α- and enantioselectivity (91:9 and 89:11, respectively). A rationale for the improvement in α selectivity in the reaction with ap-2a is that, as depicted in IX, proper alignment of the C–OMe and C–N bonds allows for effective σ_C–N→σ*_C–O hyperconjugation, leads to diminished electron density at nitrogen. This is consistent with the reduced α:γ selectivity when ap-2b was used, as the requisite conformation would exacerbate steric pressure (X). The much lower a selectivity with ap-3 (Scheme 3b) underscores the positive impact of the additional C–O bond.

Various aryl- and F₃C-substituted silyl ketimines may be converted to α-tertiary homoallylic NH₂-amines (4b-m, Table 1). In only four instances the pure α-addition isomer was not obtained (entries 1–2, 4, and 11). Reactions of hindered o-Cl and o-tolyl ketimines afforded 4c and 4e in 79% and 52% yield, respectively (entries 2 and 4). In such instances or with strongly electron-rich aryl ketimines (e.g., 1m), to maximize efficiency, in situ silyl removal was performed prior to charging the mixture with ap-2a and (Z)-3.^[15] With ketimines that are less reactive due to steric (e.g., entries 2–4) or electronic factors (e.g., entries 4 and 12), borotropic shift became more competitive and α:γ selectivity suffered.

Table 1:

Catalytic enantioselective additions to aryl-substituted ketimines.^[a]

Entry	Ar	Conv. [%][a]	α:γ[b]	Yield (pure α) [%][c]	Z:E[b]	e.r.[d]
1	oFC6H4; b[d]	>98	96:4	89[f]	98:2	98.5:1.5
2	oClC6H4; c[e]	>98	74:26	79[f]	98:2	96.5:3.5
3	oMeOC6H4; d	>98	96:4	70	98:2	>99:1
4	oMeC6H4; e[e]	>98	45:55	52[f]	98:2	95.5:4.5
5	mClC6H4; f	>98	>98:2	90	97:3	96.5:3.5
6	mF3CC6H4; g	>98	98:2	82	98:2	98:2
7	pFC6H4; h	>98	93:7	68	98:2	97:3
8	pClC6H4; i	>98	95:5	87	95:5	96:4
9	pBrC6H4; j	>98	96:4	67	97:3	98:2
10	pF3CC6H4; k	>98	98:2	61	96:4	97.5:2.5
11	pMeOC6H4; l	96	94:6	78[f]	98:2	98:2
12	pMe2NC6H4; m[e,g]	98	87:13	50	98:2	96.5:3.5

^[a]All reactions performed under N₂ atm.

^[b]Conv., d.r., and a:g ratios determined by analysis of ¹⁹F NMR spectra of unpurified mixtures; conv. (±2%) refers to disappearance of the silyl ketimine starting material.

^[c]Yield of isolated and purified material (±5%).

^[d]Enantioselectivity determined by HPLC analysis (±1%); the derived acetate was used in entry 1.

^[e]NH-Ketimine generated first.

^[f]Yield is for mixture of a- and g-addition products.

^[g]Reaction time = 8 h. See the Supporting Information for details.

Heterocyclic α-tertiary amines were prepared in high Z:E ratios and enantioselectivities, as indicated by 4n-p (Scheme 4). However, these processes presented additional challenges. Low stability of the silyl ketimine (hydrolysis to ketone) was one problem, and thus 4n, bearing a more electron donating 2-furyl substituent, was isolated in 32% yield (vs. benzofuryl- and 2-thienyl-substituted 4o-p in 77% and 84% yield, respectively). We could not obtain an indole-substituted silyl ketimine in sufficiently high purity, and attempts to promote addition to the 2- or 3-pyridyl variants were thwarted due solubility issues and/or rapid formation of the derived hemiaminal (reaction with MeOH). Efforts to carry out reactions with alkyl-substituted substrates led to the formation of hydrolyzed products (ketones), enamines, or hemiaminals (<5% desired product).

Scheme 4.

Heterocyclic, alkenyl-, and alkynyl-substituted ketimines may be used. Same conditions as in Table 1 were used, except for 4o the NH-ketimine was generated first in situ. Yield of isolated and purified material (±5%); enantioselectivity determined by HPLC analysis (±1%). See the Supporting Information for further details.

Allylic and propargylic α-tertiary amines 5 and 6 were isolated in 64% and 81% yield (α+γ isomers), respectively, and in high Z selectivity and e.r. The lower α:γ ratio (80:20) for 6 probably arises from increased stability of the hemiaminal derivative (~20% detected^[15]), owing to a more diminutive alkynyl group, which lowers the barrier to borotropic shift (see Scheme 3a).

Other Z-allyl boronates, accessible by catalytic cross-metathesis,^{[ 17 ]} may be used, providing access to products with different alkenyl moieties (Scheme 5). We prepared α-tertiary amines bearing an n-heptyl (7), a chloro (8), or a bromo (9a-b) substituent. Yields (56–91% for pure α isomers), regioselectivities (84:16–98:2 α:γ), and Z:E selectivities (≥98%) were generally high; enantioselectivity ranged from 88.5:11.5 to 97:3 e.r. Trisubstituted olefin 10 was obtained in 88% yield (pure α ), >98:2 Z:E selectivity, and 88.5:11.5 e.r.

Scheme 5.

The enantioselective approach is applicable to different allyl boronates, accessible by stereoretentive catalytic cross-metathesis. Same conditions as in Table 1 used, except that 14 h was needed for 8a-b and 10. For 8a-b and 9a-b, the silyl group was removed first. Yield of isolated and purified material (±5%); enantioselectivity determined by HPLC analysis (±1%). See the Supporting Information for further details. Abbreviation = MEM: methoxyethoxymethyl.

Another feature of the new strategy is that aminophenols can be easily prepared in multigram quantities from inexpensive starting materials,^[18] and are commercially available. By using O-benzyl-L-threonine derived ap-4, which recently became commercially available, we were able to prepare 4a in 69% yield (pure α), [97:3 Z:E and 92:8 e.r. vs. 80% yield (pure α, Scheme 6a), 98:2 Z:E, 97:3 e.r. with ap-2a]. The feasibility of synthesizing a-tertiary NH₂-amines with a halogen-substituted Z-alkene means that additional derivatives can be conveniently prepared by cross-coupling reactions, stereoretentive catalytic processes that do not require amine protection. A representative case is the two-step transformation of ketimine 1j to 11 in 71% overall yield, >98:2 Z:E selectivity, and 92.5:7.5 e.r. (Scheme 6b). Another example is a cross-coupling involving homoallylic α-tertiary NH₂-amine 9b and alkenyl–B(pin) 12, leading to the formation of 1,3-diene 13 in 54% yield (>98:2 Z/E:E/E).^[15]

Scheme 6.

The practical nature and utility of the new catalytic enantioselective method. Yield of isolated and purified material (±5%); enantioselectivity determined by HPLC analysis (±1%). See the Supporting Information for further details.

Thus, the possibility of generating products that contain a Z-alkenyl chloride, not only makes it possible to access a much wider range of other a-tertiary NH₂-amines, it can be done so with high chemoselectivity.

As noted earlier, a central attribute of the present method is that the products contain a Z alkene, transformations of which are considerably more diastereoselective than the corresponding E isomers or monosubstituted olefins.^{[9, 19 ]} The following data highlight the importance of this feature. Directed epoxide formation/regioselective cleavage with mchloroperbenzoic acid and trichloroacetic acid (Scheme 6c), according to a procedure introduced for reactions involving cyclic allylic tertiary alkylamines and γ-tertiary alkylamino α,β-unsaturated esters,^[20] afforded 14 in >98:2 regiosiomeric ratio (r.r.); mild hydrolysis converted 14 to diol 15 in 57% overall yield and 93:7 diastereomeric ratio (d.r.). The X-ray structure of amide/acetonide derivative 16 confirmed the identity of the major stereoisomer. Similarly, 10 was converted to trichloroacetate 17 in 72% yield, >98:2 r.r., and 91:9 d.r.

The gram-scale synthesis of cyclic amide 19 (1.21 g) via 18 (Scheme 6d), previously converted to BACE-1 inhibitor (Scheme 1a), further highlights the utility of the approach. A homoallylic amine with a monosubstituted alkene can also be used in the same way for this last sequence. Nevertheless, the fact that precious metal salts are not needed, the aminophenol can be prepared in significant amounts at relatively low cost, and multigram quantities of Z-crotyl– B(pin) may be accessed by simple procedures and with inexpensive starting materials,^[18] point to the truly practical nature of the approach.

Conclusion

We have described a method for enantioselective preparation of readily modifiable F₃C- and aryl-, heteroaryl-, alkynyl, or alkenyl-substituted α-tertiary Z-homoallylic NH₂-amines. Reactions involve an easy-to-handle N-silyl ketimine and a Z-allyl boronate, which might either be purchased or accessed in high stereoisomeric purity by catalytic cross-metathesis, and are promoted by a catalyst generated in situ from a readily accessible aminophenol. High α:γ selectivities arise from attractive electrostatic forces and the steric and stereoelectronic attributes of the catalyst’s L-threonine residue. Development of additional catalytic protocols for enantioselective synthesis of a broader range of α-tertiary NH₂-amines (e.g., F₃C-/alkyl-substituted) are in progress.