Driving Recursive Dehydration by p^III/p^V Catalysis: Annulation of Amines and Carboxylic Acids by Sequential C-N and C-C Bond Formation

Abstract

A method for the annulation of amines and carboxylic acids to form pharmaceutically-relevant azaheterocycles via organophosphorus P^III/P^V redox catalysis is reported. The method employs a phosphetane catalyst together with a mild bromenium oxidant and terminal hydrosilane reductant to drive successive C -N and C -C bond forming dehydration events via the serial action of a catalytic bromophosphonium intermediate. These results demonstrate the capacity of P^III/P^V redox catalysis to enable iterative redox-neutral transformations in complement to the common reductive driving force of P^III/P^V couple.

Graphical Abstract

Progress over the past decade has established the viability of the P^III/P^V=O redox couple for catalysis.^1–3 In contrast to prior notions about the kinetic inertness of the P=O bond, the incorporation of P into a ring structure can lead to swift deoxygenation by mild reagents such as hydrosilanes.^4,5 By virtue of the reducing potential of the p^III/ p^V couple, m any of these transformations are reductive in nature (Figure 1A, left).^6–8 In this vein, we have shown that four-membered ring organophosphorus catalysts effect reductive conversion of nitro⁹ and sulfonyl¹⁰ substrates (cf. Figure 1B) in which the ability to recursively renew a reactive P^III species under conditions of p^III/p^V catalysis enables a single catalyst to perform successive deoxygenative operations on a substrate (i.e. auto-tandem catalysis¹¹).

Figure 1.

(A) Complementary reductive and redox-neutral modes of P^III/p^V catalysis. (B) Catalytic heterocyclization of nitrobiaryl by recursive deoxygenation. (C) Catalytic annulation of amines and carboxylic acids by recursive dehydration.

In addition to reductive chemistry, the versatile P^III/P^V driving force can be adapted to achieve net redox-neutral transformations when paired with an appropriate oxidant as evoked in Mukaiyama’s conceptualization of an “oxidation-reduction condensation.”¹² Within a P^III/P^V-catalytic context,¹³ the introduction of a mild chemoselective halenium oxidant, for instance, can shunt the reductive manifold into a net redox-neutral mode where the key reactive catalytic intermediate is not a phosphine but rather a halophosphonium cation^14,15 (Figure 1A, right). Indeed, halophosphonium intermediates have been invoked by Rutjes and van Delft⁵ and Mecinović¹⁶ in the context of P^III/P^V-catalyzed Appel halogenation and N-acylation reactions, respectively.¹⁷ In view of the fact that phosphonium reagents have been described as having “virtually ideal properties as selective oxygen extractors for net dehydration reactions,”^18,19 the potential to achieve recursive dehydrations in an auto-tandem catalytic manner via a net redox-neutral mode of p^III/p^V catalysis could be expected to present new opportunities for serial bond formation.

We show here an annulation of amines and carboxylic acids via recursive dehydration driven by a redox-active organophosphorus catalyst cycling in the P^III/P^V couple (Figure 1C). This auto-tandem catalytic system enables the elaboration of simple, commercially available starting materials into pharmaceutically- relevant azaheterocycles through a condensation/cyclodehydration sequence in a one-pot catalytic protocol. The success of the approach relies on the mutual compatibility and functional interplay of the reducing and oxidizing reagents with the organophosphorus catalyst in order to orchestrate a sequence of distinct C -N and C -C bond-forming events. The ability of P^III/P^V redox catalysis to encompass such recursive dehydration stands as a complement to existing deoxygenation methods, thus broadening the scope of transformations accessible to this catalytic mode.

To evaluate the possibility of recursive dehydration driven by P^III/P^V redox cycling, the tandem amidation/cyclodehydration of amine 1a and carboxylic acid 2a to generate pyrroloquinoxaline 3a was evaluated (Table 1). Optim al conditions using 1,2,2,3,4,4- hexamethylphosphetane P-oxide 4• [O]²⁰ as catalyst, diethyl bromomalonate (DEBM ) as oxidant, and phenylsilane as terminal reductant yielded the desired product in 94% yield, isolable on 0.4 mmol scale in 84% yield (Table 1, entry 1). A mild, weakly oxidizing bromenium reagent was found to be essential for the redox compatibility of the system , as demonstrated by Rutjes and van Delft;⁵ the related diethyl (methyl)bromomalonate (DEMBM) was similarly competent in the transformation (entry 2), but the more strongly-oxidizing N -bromosuccinimide resulted in poor conversion to product (entry 3). Chlorenium oxidants such as diethyl chloromalonate (DECM) and carbon tetrachloride gave no dehydrative heterocyclization (entries 4 and 5); instead, am ide 3a’ was obtained in 70% yield, indicating chlorophosphonium ion com petency in C -N forming amidation but not C -C forming cyclodehydration.²¹

Table 1.

Discovery and Optimization of Phosphacatalytic Iterative Condensation/Annulation of Amine and Carboxylic Acid

Entry	R3P=O	X+ source	Yield (%)a
1	4•[O]	DEBM	94 (84)b
2	4•[O]	DEMBM	90
3	4•[O]	NBS	10
4	4•[O]	DECM	0
5	4•[O]	CCl4	0
6	5•[O]	DEBM	26
7	6•[O]	DEBM	46
8	7•[O]	DEBM	50
9	8•[O] or 9•[O]	DEBM	0
10	none	DEBM	0
11	4•[O]	none	0
12	4•[O], no PhSiH3	DEBM	0
13	4	DEBM	90
14	[4•Br]Br	DEBM	87

^aYields determined through ¹H NMR analysis with the aid of an internal standard. See Supporting Information for full synthetic details and yields of intermediate amide 3a’.

^bIsolated yield on 0.4 mmol scale. DCE = 1,2-dichloroethane; DEBM = diethyl bromomalonate; DEMBM = diethyl (methyl)bromomalonate; NBS = N-bromosuccinimide; DECM = diethyl chloromalonate.

With respect to catalyst, variation of the phosphetane exocyclic moiety to ethyl (5•[O], entry 6), phenyl (6• [O], entry 7), or pyrrolidino (7• [O], entry 8) all resulted in substantial decrease in the efficiency of the reaction, while the use of acylic phosphine oxides (8•[O] or 9• [O], entry 9) fail to promote the cyclocondensation (see SI). Control experiments confirm that no azaheterocycle product is observed in the absence of any of 4•[O], phenylsilane, or DEBM (entries 10–12).²² Furthermore, employing P^III species 4 or pregenerated bromophosphonium [4•Br]Br²³ in place of phosphine oxide 4 •[O] resulted in com parable efficiency, consistent with the notion of P^III/P^V=O redox cycling (entries 13–14).

The optimized phosphacatalytic protocol provides direct access to complex azaheterocycles from simple and readily available amine and carboxylic acid starting materials (Figure 2). A variety of carboxylic acids are efficiently incorporated into pyr-roloquinoxalines, including those possessing olefinic and aryl functionalities (3c-3i, 54–89% yields). This protocol was also readily translated to larger scale reactions, as a 5 mmol scale reaction of 1-(2-am inophenyl)pyrrole and butyric acid provided 1.04 g of com pound 3b in 99% yield with 8 mol% loading of organophosphorus catalyst 4 •[O]. As demonstrated by products 3d- 3i, the reaction efficiency is relatively independent of substitution on benzoic acid coupling partners, including changing steric profile (54–89% yields). Critically, acids containing polar functionalities, including alkyl ethers, thioethers, sulfonamides, and alkyl halides, undergo efficient iterative dehydration to provide heterocycles in good to excellent yields (3a, 3j-3p, 41–90% yields). Of particular note are the amino acid-incorporating products 3k and 3p, which originate from protected Ts-Gly-OH and Ts-Phe-OH. Further, both primary and secondary haloalkane functionalities are conserved under this P^III/P^V-catalytic m anifold (3l and 3m, 93% and 98% yields, respectively). Substitution on the aniline ring was well- tolerated; both ortho- and meta-substituted pyrroloanilines were readily incorporated into heterocyclic scaffolds (3n and 3o, 86% and 95% yields, respectively). Chiral carboxylic acids, such as ibuprofen and naproxen, are incorporated with good yield and high stereochemical fidelity (3q, 67 %, 95:5 e.r.; 3r, 68%, 92.5:7.5 e.r.) under modified recursive dehydration conditions (precatalyst [4•Br]Br, MeCN, 50 °C; see Supporting Information for full synthetic details).

Figure 2.

Examples of auto-tandem phosphacatalytic annulation of amines and carboxylic acids. All yields isolated on 0.4 mmol scale unless indicated otherwise. See Supporting Information for full synthetic details. ^aIsolated yield on 5.0 mmol scale, with 8 mol% of 4•[O]. ^bReaction conducted with Ph₂SiH₂ (4.4 equiv) in MeCN. ^cReaction conducted at 60 °C for 20 h. ^dReaction conducted on 0.2 mmol scale using 15 mol% of [4•Br]Br in MeCN at 50°C for 20–40 h. ^eYield determined by ¹H NMR with internal standard. ^fReaction conducted with DEMBM. DEBM = diethyl bromomalonate; DEMBM = diethyl(methyl)bromomalonate; DCE = 1,2-dichloroethane; Bn = benzyl; Ts = tosyl; ⁱBu = iso-butyl.

When N-alkyl amine substrates were initially employed under standard conditions, an undesired byproduct arising from N-alkylation by DEBM was identified by GCMS. However, replacing DEBM with its methyl-substituted analogue, DEMBM, abated this deleterious pathway, restoring the high degree of redox compatibility necessary for the catalytic system. Consequently, o-pyrrolobenzylamine could be efficiently coupled with carboxylic acids via iterative dehydration to provide the corresponding pyrrolo-benzodiazepines, a prevalent bioactive scaffold,²⁴ in good yields (3s-3u, 66–86% yields).²⁵ Heterocycles 3t and 3u, compounds investigated by Janssen for their antifungal activity,²⁶ could be prepared in a single synthetic operation in 86% and 73% yields, respectively. Further, tryptamines and phenethylamines could be transformed into dihydro-β-carboline and dihydroisoquinoline products in good yields (3v-3x, 70–77% yields).²⁷ Notably, both of these scaffolds are found in bioactive pharmaceutical agents and natural products,²⁸ as demonstrated by the assembly in a single step of dihydroisoquinoline natural product 3,4- dihydropapaverine from commercial starting materials with good efficiency (3x, 70% yield).

Concerning the catalytic mechanism, in situ¹H NMR spectroscopy revealed a rapid initial conversion of reactants 1a and 2a into am ide intermediate 3a’, followed by comparatively slow formation of heterocycle 3a (see SI, Sect. VI), establishing a stepwise reaction sequence for the auto-tandem catalytic process in which C -C bond-forming cyclodehydration is kinetically limiting. Despite the observation that 4, 4•[O], or 4•Br⁺ are each competent precatalysts (vide supra), in situ³¹P NMR and DART- MS analyses show that none of these com pounds represent the catalytic resting state. Rather, experiments are most consistent with resting state B1, an adduct of the phosphacyclic catalyst and am ide A1 (Figure 3A). Indeed, independent reaction of [4•Br]Br with A1 gives rise to spectroscopic signals indistinguishable from those observed under the catalytic steady state, and this species was shown to lead to C -C bond-forming cyclodehydration.²⁹ Furthermore, spectroscopically indistinguishable species are observed by³¹P NMR spectroscopy when either N-methylacetamide and N,N-dimethylacetamide are introduced in lieu of reactive am ides to a mixture containing catalytic components (i.e. 4•[O], PhSiH³, DEBM).

Figure 3.

(A) ³¹P NMR studies. (B) Proposed mechanism of auto-tandem catalytic dehydrative annulation of amines and carboxylic acids. Methyl groups excluded from 4 for clarity. DEBM = diethyl bromomalonate.

Figure 3B depicts a plausible auto-tandem catalytic reaction mechanism consistent with the foregoing experimental data. From phosphine oxide 4•[O] as precatalyst, entry to the C -N bond-forming cycle (Figure 3B) is initiated by kinetically-facile phenylsilane-mediated reduction to phosphine 4, followed by rapid halophilic reaction³⁰ with DEBM leading to bromophosphonium ion 4•Br+. Bromophosphonium cation 4•Br+ effects intermolecular amidation between acid 1 and amine 2, presumably via intermediate C in analogy to established precedent for amine N-acylation by activated acyloxyphosphoniums, thereby returning phosphine oxide 4•[O]. In C -C bond-forming second phase, phosphonium ion 4•Br⁺ is again generated by a reduction- oxidation sequence with PhSiH₃ and DEBM, respectively. Exchange of bromide for the am ide substrate A then leads to activated species B, which is assigned as the catalytic resting state. Cyclization ensues to provide the product 3, liberating phosphine oxide 4•[O] and closing the catalytic cycle.³¹ The two noteworthy conclusions emerging from this mechanistic picture are: (1) Turnover of phosphine oxide 4•[O] to phosphine 4 is not kinetically limiting; and (2) The concentration of reducing phosphine 4 remains negligibly low during catalysis as a function of the efficient reaction with the oxidative halenium shunt.

The net redox neutral character of the recursive dehydration (and the absence of appreciable concentrations of phosphine 4) enables chemoselective annulation in preference to established P^III/P^V-catalyzed reductive transformations. This orthogonality is illustrated in the context of p-nitrohydrocinnam ic acid (10), possessing both carboxylic acid and nitroarene moieties (Figure 4). Under the recursive dehydration conditions described herein, catalyst 4•[O] drives the selective annulation of carboxylic acid 10 to yield pyrroloquinoxaline 11 in 97% yield. With the same catalyst (4•[O]) but omission of the DEBM shunt, the nitro group is then reductively addressed to effect intermolecular C -N cross coupling^9c and yield fully deoxygenated species 12 in 72% yield. This sequence, in which a single organophosphorus catalyst executes four distinct oxygen excisions, demonstrates the complementarity of the reductive and redox-neutral P^III/P^V-catalytic manifolds owing to the disparate reactivity of the phosphetane in its different oxidation states as a function of the addition or exclusion of an exogenous oxidant.

Figure 4.

Selective functionalization of carboxylic acid- and nitro-containing substrate via sequential redox-neutral recursive dehydration, then reductive recursive deoxygenation, using a single catalyst 4•[O]. Reaction conditions: (a) 1a (1.0 equiv), 10 (1.05 equiv), DEBM (2.4 equiv), PhSiH (2.2 equiv), 4•[O] (15 mol%), DCE, 80 °C; (b) 11 (1.0 equiv), 4-MeO-C₆H₄ -B(OH)₂ (1.1 equiv), PhSiH₃ (2.0 equiv), 4•[O] (15 mol%), mxylene, 120 °C.

In conclusion, we have demonstrated that a small-ring phosphetane catalyst can induce iterative dehydrative C -N and C -C bond-forming reactions, enabling direct azaheterocycle synthesis from carboxylic acids and amines via recursive dehydration. Through the synergistic use of mild hydrosilane reductant and bromenium oxidant, the elements of water can be catalytically removed in the form of an O-atom and two protons with complete redox compatibility. We anticipate that this phosphacatalytic dehydration manifold will prove generally enabling for the redox-neutral functionalization of oxygenated organic functionalities to accomplish C -C and C-heteroatom bond forming events via condensation, especially in a recursive fashion.