Preparation of N-Aryl Amides by Epimerization-Free Umpolung Amide Synthesis

Abstract

Amide synthesis is one of the most widely practiced chemical reactions, owing to its use in drug development and peptide synthesis. Despite the importance of these applications, the attendant effort to eliminate waste associated with these protocols has met with limited success, and pernicious α-epimerization is most often minimized but not eliminated when targeting challenging amides (e.g. N-aryl amides). This effort has focused on what is essentially a single paradigm in amide formation wherein an electrophilic acyl donor reacts with a nucleophilic amine. Umpolung Amide Synthesis (UmAS) emerged from α-halo nitroalkane reactions with amines and has since been developed into a method for the synthesis of enantiopure amides using entirely catalytic, enantioselective synthesis. However, its inability to forge N-aryl amides has been a longstanding problem, one limiting its application more broadly in drug development where α-chiral N-aryl amides are increasingly common. We report here the reaction of α-fluoronitroalkanes and N-aryl hydroxyl amines for the direct synthesis of N-aryl amides using a simple Brønsted base as the promoter. No other activating agents are required, and experiments guided by mechanistic hypotheses outline a mechanism based on the UmAS paradigm and confirm that the N-aryl amide, not the N-aryl hydroxamic acid, is the direct product. Ultimately, select chiral α-amino-N-aryl amides were prepared with complete conservation of enantioenrichment, in contrast to a parallel demonstration of their ability to epimerize using the conventional amide synthesis alternative.

Graphical Abstract

Amide synthesis is arguably one of the most broadly- and frequently-used organic reactions, stimulating a massive infrastructure development that includes technology (e.g. solid-phase peptide synthesis)¹ and a broad array of raw materials (e.g. commercially-available carboxylic acids and amines).² The protocols practiced for amide synthesis are economically unsustainable,^3,4,5,6 and they are increasingly associated with health hazards that include human allergic response to the coupling reagents.⁷ At research scale where external safety measures may derisk the use of coupling agents and nucleophilic promoters, the overall strategy to prepare α-chiral amides is not only step-intensive, but also susceptible to epimerization.^8,9,10 Armed with the belief that improvements within this paradigm may be only marginal, we have developed Umpolung Amide Synthesis (UmAS) into a method that can alleviate many of these issues, providing a means to prepare complex amides using entirely catalytic, enantioselective tactics.^{11,12,13,14,15} A key limitation, however, is the fact that aryl amines (anilines) have been uniformly unsuccessful in UmAS. We now address this shortcoming by the discovery that N-aryl hydroxyl amines are suitable acceptors in UmAS, unexpectedly leading directly to N-aryl amides instead of hydroxamic acids.^16,17

Recent reports of the use of activated hydroxyl amines as electrophilic nitrogen sources led us to investigate the reactivity of α-halo nitronate nucleophiles with N-aryl hydroxyl amines and their activated derivatives (Figure 1).^18,19 These efforts were largely unsuccessful with vanishingly small amounts of the desired N-aryl amide forming in the best cases. Within this dataset, use of an α-bromo nitroalkane (1a), the most common type of UmAS donor, with N-aryl hydroxyl amine 2 gave only 1% and 16% yield of the desired N-aryl amide (Table 1, entries 1–2). Although base selection can dramatically improve the yield of UmAS, the investigation of alternatives was met with limited improvement (Table 1, entries 3–4), but nitronate formation was clearly necessary (Table 1, entry 5). Cesium carbonate consistently outperformed alternative bases, so it was selected for subsequent studies.²⁰ A lower concentration aided stirring since the carbonate base was minimally soluble, and this had little effect otherwise on the reaction outcome (Table 1, entry 6 vs. 7). Reasoning that the reactivity of the α-halonitronate might be better matched to the unusual N-aryl electrophile, we compared the reactivity of α-bromo nitroalkane 1a to α-fluoro nitroalkane 1b (Table 1, entries 7–8). This experiment could be performed at room temperature in both cases but the yield of the desired N-aryl amide was doubled when using the α-fluoronitroalkane. Further optimization of the protocol led to a 76% yield from 1b using three equivalents of cesium carbonate at room temperature (Table 1, entry 9). Throughout these experiments, formation of azoxy 4 from the hydroxyl amine substrate was noted.²¹

Figure 1.

Comparison of conventional amide synthesis (top) and initial UmAS approach to N-aryl amide synthesis (bottom).

Table 1.

Reaction optimization for α-halo nitroalkane conversion to N-aryl amide.^a

entry	X	1	base	temp (°C)	yield (%)
1	Br	a	TEA	60	1
2	Br	a	K2CO3	60	16
3	Br	a	KOH	60	13
4	Br	a	Cs2CO3	60	30
5	Br	a	none	60	0
6	Br	a	Cs2CO3	rt	33
7b	Br	a	Cs2CO3	rt	30
8b	F	b	Cs2CO3	rt	64
9b,c	F	b	Cs2CO3	rt	76

^aGeneral procedure: nitroalkane (1 equiv.), hydroxylamine (1.5 equiv.), and base (2 equiv) in toluene (0.1 M) were sparged with Ar for ~2 min. The reaction was stirred for 16–26 hours.

Yields measured by ¹H NMR relative to CH₂Br₂.

^b0.05 M.

^c3 equiv. base used.

The initial limits of N-aryl amide synthesis from a common α-fluoro nitroalkane (1b) and a variety of N-aryl hydroxyl amines were explored as summarized in Table 2. Electron-rich N-aryl amides were of particular interest, and the most common substituents (-NH₂ (7b), -OMe (7e)) were tolerated. Although double acylation was not expected based on our mechanistic hypothesis, we confirmed that no double acylation product could be detected for 7b. Common electron-deficient N-aryl amides (7f, 7h, 7i) were also prepared in moderate to good yield (Table 2). A variety of α-fluoro nitroalkane donors were also prepared and compared using a common N-aryl hydroxyl amine (2). These included aryl (7k-l, q-s) and alkyl amides (7m-p, t-u).

Table 2.

N-Aryl amide synthesis from α-fluoronitroalkane and N-aryl hydroxyl amine.^a

^aGeneral procedure: The aryl hydroxylamine (1.5 equiv), Cs₂CO₃ (3 equiv), and fluoronitroalkane were stirred in solvent (0.05 M) in a vial for 24 h.

Yields were measured by ¹H NMR (relative to CH₂Br₂) after filtration through a pad of silica gel. Isolated yields (after chromatography for most compounds) are reported in the SI.

Chemoselective amidations are particularly useful, and this method provides the amide derived exclusively from N-hydroxyl amines (Scheme 1, 8→9, 10→11). Finally, some reactions of cyclopropyl-substituted bromonitromethane are not stereospecific,²² but use of α-fluoro nitronate 5a in these studies led to N-aryl amide 7o as a single diastereomer. Intramolecular cyclization by Michael addition followed N-aryl amide formation in the case leading to 7u due to the electrophilic enoate functionality.

Scheme 1.

Umpolung N-aryl amide synthesis: regio- and diastereospecific reactions.^a

^a Reaction conditions mirrored those for Table 2. See SI for complete details. Isolated yield shown for 7o.

Despite the welcome simplicity of the reaction, using only a common base to promote the coupling, we were mystified by the observation that attempts to further activate the hydroxyl amine by tosylation, or acylation at oxygen (e.g. Figure 1), led only to recovery of starting materials. A formal substitution at nitrogen would lead directly to the amide products observed. Ultimately, our mechanistic hypotheses for the experiments in Table 1 were based on the oxidation of the hydroxyl amine to an aryl nitroso, but this redox pathway should lead to an N-aryl hydroxamic acid product instead of N-aryl amide.16

Initial mechanistic investigations therefore first reconsidered the non-umpolung paradigm for amide synthesis. Namely, if the N-aryl hydroxylamine is functioning as a nucleophile, as either a hydroxylamine N-nucleophile, or precursor to its aniline derivative, then these conditions should apply successfully to an aniline (Scheme 2). The competition between N-aryl hyroxylamine 2 and para-fluoro aniline (12) with α-fluoronitroalkane 1b (1:1:1 ratio) was found to produce solely the hydroxylamine-derived N-aryl amide (3) in 71% yield (Scheme 2, eq 2), consistent with the experiments in Scheme 1.²³ A similar competition experiment with a more nucleophilic amine was also examined, leading to a similar result and a trace amount of benzyl amine-derived amide 15 (3%, Scheme 2, eq 3). A series of experiments using possible hydroxylamine-derived products as reactants led to the knowledge that azoxy 4 and nitroarene (para-chloro nitrobenzene)²⁴ are not competent precursors to amide. Collectively, these experiments directed our hypothesis-driven mechanistic experiments back to an umpolung mechanism.

Scheme 2.

Experiments probing the initial mechanistic hypothesis.^a

^a Reaction conditions generally mirrored those for Table 2. Yields were measured by ¹H NMR (relative to CH₂Br₂) after filtration through a pad of silica gel.

A working hypothesis consistent with these observations is summarized in Scheme 3. The N-aryl hydroxylamine is first oxidized to an aryl nitroso (20). This electrophile reacts with the α-fluoronitronate (21) to form haloamino nitroalkane (HANA)²⁵ 22 which then collapses to fluoronitrone 23. A new hydroxyl amine (19) then converts 23 to amide product, and consequently, forms a new aryl nitroso (20). An interesting feature of the catalytic cycle in Scheme 3 is the requirement for an aryl nitroso intermediate (20), but a parallel need for N-aryl hydroxyl amine (19) to effect the final step from 23 to amide 24. Hence, substitution of the hydroxyl amine by an aryl nitroso should arrest the sequence at 23, owing to the removal of the N-aryl hydroxyl amine and its putative role as a key reductant. In the event (Scheme 2, eq 4), use of nitrosobenzene provided traces of amide 7d alongside a major product assigned the structure of fluoronitrone 16 as a 77:23 mixture of geometric isomers. This species was unstable, but immediate analysis was revealing by ¹H NMR (pair of doublets at 3.78, 3.61 ppm) and ¹⁹F NMR (singlets at −26.4, −27.4 ppm), leading to our assignment (16). Additional experiments established the coupling of fluorine with proton. Although no direct analogies for 16 could be found in the literature, it has been postulated previously^26,27 and other α-heteroatom nitrones are known.^28,29 HNO₂ elimination from a HANA is analogous to formation of nitrones from nitroalkane addition to nitrosoarenes³⁰ and azomethine ylide formation from α-fluoro nitroalkane addition to azodicarboxylate.²⁵

Scheme 3.

Umpolung N-aryl amide synthesis: mechanistic hypothesis.

The mechanistic hypothesis in Scheme 3 suggests that when the putative fluoro nitrone (23) is formed stoichiometrically from an aryl nitroso, a separate N-aryl hydroxyl amine should facilitate its conversion to amide product by acting as the reductant. Indeed, this was the outcome when using reactants with distinct aryl groups (Scheme 2, eq 5). A similar crossover experiment was performed to establish that a discrete N-aryl hydroxamic acid is not formed during the reaction (Scheme 2, eq 6). The interdependency of nitrosoarene formation and fluoronitrone conversion to amide (23+19→24) is an essential component of this ‘direct’ umpolung N-aryl amide synthesis.

An umpolung mechanism for N-aryl amide synthesis from N-aryl hydroxyl amines provides a unique opportunity to target α-chiral, highly acidic N-aryl amides. Aryl glycines are more susceptible to epimerization relative to other α-amino acids,^31,32 with the rate of phenyl glycine epimerization measured at 9-fold higher than that of alanine. We recapitulated this behavior with an N-aryl amide of an N-aryl glycine and then designed experiments to compare the degree of conservation of configurational integrity during its synthesis. Scheme 4 describes the results when preparing 26 from 25 (90% ee) using EDC/HOAt under basic conditions in DMF. The desired N-aryl amide is obtained in 57% yield, but only 65% ee (Scheme 4, eq 7). Similarly, the product was prepared in an improved 82% ee and 83% yield when using DEPBT (eq 8), a coupling reagent known to be effective with epimerization-prone amides.³³ By comparison, α-fluoronitroalkane 27 was deployed as a mixture of diastereomers (1:1) homochiral at the benzylic amine carbon (98% ee for each diastereomer),^34,35 and when stirred with N-para-chlorophenyl hydroxylamine, the desired amide 26 was formed with complete conservation of ee (eq 9). Although the yield (41%) requires further optimization, the outcome is consistent with the mechanism of UmAS wherein no electrophilic intermediate is formed (Scheme 3). We noted, however, that extended exposure of the product to the basic conditions can result in racemization of 26 (t_1/2 ~ 220 h, toluene, 0 °C, 98→75% ee at 105 h). This further illustrates the heightened acidity of N-aryl amides, and the challenge they present for synthesis.

Scheme 4.

Comparison between conventional amide synthesis (EDC, DEPBT) and UmAS of an N-Aryl-α-Aryl glycinamide target to assess conservation of enantiomeric excess (% cee).

^a See SI for complete details. Reaction conditions generally mirrored those for Table 2. Yields were measured by ¹H NMR (relative to CH₂Br₂) after filtration through a pad of silica gel.

Incorporation of this umpolung N-aryl amide synthesis into synthesis planning was targeted next. N-Aryl amide 28³⁶ is a member of the Globalagliatin-class of glucokinase activators for the treatment of diabetes.³⁷ We envisioned an approach to 28 that involved the two key disconnections illustrated in Scheme 5, requiring chemistry that allows an inverted ketene synthon.^38,39 Friedel-Crafts acylation of 29 provided aryl ketone 31, although the reaction was somewhat dependent on the age of the aluminum chloride used. Oxidation to the sulfone, and Wittig methylenation delivered terminal alkene 32. Nitration prepared the substrate for enantioselective reduction using 34. The product was obtained in good yield and encouraging enantioselection (86% ee) using a new enantioselective conjugate reduction.⁴⁰ Fluorination readied 36 for N-aryl amidation using 2-pyridyl hydroxyl amine. Complete conservation of enantiomeric enrichment was observed. Forging the key aryl amide bond was met first with some disappointment, as 28 was formed in only 26% yield. However, experimentation revealed that copper(II) acetate enhanced the amidation, providing 28 in 69% yield. This discovery was based on the hypothesis that a Lewis acidic and potentially redox-active metal might affect the rate of HANA formation and/or turnover.16d Overall, the combination of nitroalkene reduction and UmAS with arylated hydroxyl amine realized the inverted ketene character for the first time in an N-aryl amide synthesis.

Scheme 5.

Development of a double umpolung approach using N-Aryl UmAS to a glucokinase activator.

In conclusion, we have developed an umpolung amide synthesis that forms N-aryl amides directly. Anilines have been uniformly unsuccessful amines in UmAS since its introduction, but now N-aryl hydroxyl amines enable a remarkably simple protocol, requiring only base to access N-aryl amides. The overall simplicity, however, belies a more complex hydroxylamine-nitroso redox process. The redox step is coupled to ‘turnover’ of a putative α-fluoro nitrone intermediate, a mechanistic hypothesis supported by a study that outlines the involvement of key intermediates while establishing that the expected hydroxamic acid product is not formed or otherwise involved. Central to the UmAS approach is the avoidance of an epimerization-prone electrophilic acyl donor. UmAS with an N-aryl hydroxyl amine provided the N-aryl amide product directly, and without epimerization, unlike two leading coupling reagents used in conventional amide synthesis where the lower nucleophilicity of anilines is known to be problematic for yield and epimerization. Future studies will target other unsolved challenges in conventional amide synthesis while further probing the mechanistic details of HANA intermediates formed from nitroso electrophiles.