Imine Directed Cp*Rh(III)-Catalyzed N–H Functionalization and Annulation with Amino Amides, Aldehydes, and Diazo Compounds

Abstract

A multicomponent annulation that proceeds by imine directed Cp*Rh(III)-catalyzed N–H functionalization is disclosed. The transformation affords piperazinones displaying a range of functionality and is the first example of transition metal-catalyzed multicomponent N–H functionalization. A broad range of readily available α-amino amides, including those derived from glycine, α-substituted, and α,α-disubstituted amino acids, were effective inputs and enabled the incorporation of a variety of amino acid side chains with minimal racemization. Branched and unbranched alkyl aldehydes and various stabilized diazo compounds were also efficient reactants. The piperazinone products were further modified through efficient transformations. Mechanistic studies, including X-ray crystallographic characterization of a catalytically competent five-membered rhodacycle with imine and amide nitrogen chelation, provide support for the proposed mechanism.

Graphical Abstract

A multicomponent, imine directed Cp*Rh(III)-catalyzed N–H functionalization/annulation is reported. This method represents the first example of multicomponent N–H functionalization and allows for rapid access to a heterocyclic product. Further transformations and a plausible mechanism based on X-ray crystallographic characterization of a catalytically competent rhodacycle intermediate are disclosed.

Nitrogen heterocycles appear abundantly in drugs and clinical candidates.^[1] For this reason, methods for the rapid synthesis of heterocycles are valued with a premium placed on accessing nonaromatic heterocycles capable of three-dimensional display of functionality.^[2] Multicomponent reactions provide a powerful approach to efficiently access complex structures from simple inputs.^[3] These processes are advantageous with regard to their step-economy, facile setup, and modularity of starting materials for introducing structural diversity. Important and well-known multicomponent reactions enable the synthesis of nonaromatic nitrogen heterocycles. The Hantzsch dihydropyridine synthesis was one of the earliest to be reported and has been used for the discovery and development of drugs (Scheme 1a).^[4] In this process, 1,4-dihydropyridines are formed from the reaction between an aldehyde, two equivalents of a β-keto ester, and ammonia. Another example is the Povarov reaction for the synthesis of tetrahydroquinolines (Scheme 1b).^[5] This multicomponent reaction leverages the rapid in situ condensation of an aldehyde and aniline to form an iminium ion intermediate under acidic conditions, which undergoes an aza Diels-Alder reaction to form the tetrahydroquinoline product. Powerful methods for the synthesis of five-membered heterocycles have also been developed utilizing 1,3-dipolar cycloadditions, most notably by azomethine cycloaddition of imines, often prepared in situ, to provide pyrrolidines (Scheme 1c).^[6]

Scheme 1.

Multicomponent syntheses of heterocycles.

In contemplating the design of new multicomponent reactions for nonaromatic nitrogen heterocycle synthesis, the nature of the starting inputs is a central consideration because it dictates the extent of diverse functionality that can readily be introduced. All the multicomponent reactions depicted in Scheme 1a–c rely on aldehyde reactants, which are desirable inputs due to their widespread commercial availability and facility for imine formation. Moreover, as exemplified by azomethine ylide cycloaddition (Scheme 1c), we were inspired to use amino acid derivatives as starting inputs given their biological importance, ready availability, and synthetic utility.^[7]

Herein, we report a new multicomponent approach to prepare six-membered nitrogen heterocycles as demonstrated for piperazinones, an increasingly used ring system in pharmaceutical candidates.^[8] This approach relies on transition metal-catalyzed imine directed amide N–H functionalization, a process that to our knowledge is unprecedented.^[9] Moreover, while N–H functionalization of amides has been demonstrated under Rh(II),^[10] Ru(II),^[11] and Ir(III)^[12] catalysis, this method provides the first example of Rh(III)-catalyzed N–H functionalization. The transformation proceeds by three-component coupling of α-amino amides, aldehydes, and diazo compounds (Scheme 1d). Despite the propensity for the imine intermediate, formed by in situ condensation of the amino amide and aldehyde, to cyclize to give an imidazolidinone,^[13] this undesired pathway occurs only minimally under the optimized reaction conditions. Short reaction times with microwave (MW) heating further facilitate the rapid preparation of piperazinones. This method exhibits good functional group compatibility with diverse and easily accessed starting materials. Notably, we have obtained X-ray crystallographic data for a catalytically competent rhodacycle of the key imine intermediate, which, along with additional mechanistic studies, provides support for a plausible catalytic cycle.

We began our investigation by exploration of the key reaction parameters using N-methyl glycinamide (1a), isovaleraldehyde (2a), and diethyl diazomalonate (3a) to form piperazinone 4a (Table 1 and Tables S1–S4). While Rh(II) catalysts are commonly used for N–H functionalization with diazo compounds,^[10] Rh₂(OAc)₄ gave very poor conversion to the desired product. Cp*Rh(III) catalysts have not previously been reported for N–H functionalization with diazo compounds, but these catalysts are well documented for directed C–H functionalization for the synthesis of nitrogen heterocycles by in situ imine formation.^[14] We therefore explored whether these catalysts could mediate imine directed N–H functionalization. A variety of conditions were explored with the preformed cationic [Cp*Rh(MeCN)₃](SbF₆)₂ catalyst. The use of K₂CO₃ or KHCO₃ in the presence of 3 Å molecular sieves (MS) with microwave heating at 100 °C for 1 h were optimal, providing piperazinone 4a in good yield. Pivalic acid as an additive gave a significantly lower product yield with competitive cyclization of the imine intermediate to form the imidazolidinone, which is known to be facilitated under acidic conditions.^[13] Analogous cationic group IX metals, namely Cp*Co(III) and Cp*Ir(III), were ineffective catalysts for this reaction. Conventional heating at 60 °C for 16 h gave a comparable yield to that observed under microwave conditions, which is advantageous for larger scale reactions as demonstrated by the isolation of piperazinone 4a in 71% yield when performed on a 3 mmol scale (Table 1).

Table 1.

Amino amide side chain scope.^[a]

^[a]Conditions: 1 (0.2 mmol), 2a (0.4 mmol), and 3a (0.3 mmol). Isolated yields are reported.

^[b]3.0 mmol scale at 60 °C (conventional heating) for 16 h.

^[c]120 °C (MW), 0.2 M.

^[d]KHCO₃ instead of K₂CO₃.

Having established the optimal reaction conditions, we first explored the scope for amino amides (Table 1). In addition to piperazinone 4a prepared from N-methyl glycinamide, α,α-disubstituted amino amides provided piperazinones 4b and 4c in good yields, although a higher temperature and concentration were employed for the preparation of the more sterically hindered spirocyclic derivative 4c. Enantiomerically pure monosubstituted amino amides enabled the incorporation of a second stereogenic center in products 4d to 4j, with each product obtained in high yields with moderate levels of diastereoselectivity. The piperazinones derived from the two enantiomers of N-methyl phenylalaninamide, 4d and 4e, were used to assess whether racemization had occurred under the reaction conditions. While 4d was obtained in 92:8 er with K₂CO₃ as the base, racemization was suppressed when the less basic KHCO₃ was used such that 4d was produced in comparable yield and with 97:3 er (see the Supporting Information). For this reason, KHCO₃ was used as the base for all examples with monosubstituted amino amides. Separation of the diastereomers of 4d by column chromatography enabled rigorous determination of the relative stereochemistry of the major diastereomer of 4d by X-ray crystallography (see the Supporting Information).^[15]

Diverse amino amide starting inputs were generated from readily accessible amino acid derivatives. In addition to 4d and 4e prepared from the phenylalanine-derived amino amides, a simple methyl group could be installed in 4f as well as the more sterically encumbered isopropyl group in 4g. Ether (4h) and carbamate (4i) functional groups were also compatible with the reaction conditions. Importantly, 4j was formed in good yield from a tryptophan-derived amino amide without protection of the indole nitrogen.

A variety of aldehydes were also explored (Table 2). In addition to β-branched isovaleraldehyde used to prepare 4a to 4j, cyclic and acyclic α-branched aldehydes provided 4k to 4o in good yields. The strained and pharmaceutically relevant cyclopropyl group^[16] (4l and 4m) as well as heterocyclic substituents (4n and 4o) were also installed from the aldehyde input. Furthermore, unbranched aldehydes were used to introduce pendent phenyl (4p), ether (4q), and ester (4r) groups. Aromatic aldehydes were ineffective coupling partners due to competitive Cp*Rh(III)-catalyzed ortho C–H functionalization of the benzaldimine intermediate.

Table 2.

Aldehyde scope.^[a]

^[a]Conditions: 1 (0.2 mmol), 2 (0.4 mmol), and 3a (0.3 mmol). Isolated yields are reported.

^[b]120 °C (MW), 0.2 M.

Next, we evaluated the scope of stabilized diazo compounds (Table 3). Aside from diethyl diazomalonate (see Tables 1 and 2), dimethyl diazomalonate and di-tert-butyl diazomalonate were used to provide 4s to 4v. However, diazo Meldrum’s acid did not provide any product, and an aryl diazoacetate afforded 4w, albeit in low yield. By employing a diazo compound derived from fluorenone, the spirocyclic piperazinone 4x was prepared. Moreover, diazo compounds derived from substituted isatins provided access to a different class of spirocyclic derivatives (4y and 4z), with the trityl substituted isatin-derived diazo compound providing 4z with good diastereoselectivity (87:13 dr). The relative stereochemistry of the major diastereomer of 4y was determined by X-ray crystallography (see the Supporting Information).^[15]

Table 3.

Diazo compound scope.^[a]

^[a]Conditions: 1 (0.2 mmol), 2 (0.4 mmol), and 3 (0.3 mmol). Isolated yields are reported.

^[b]KHCO₃ in place of K₂CO₃.

^[c]0.3 mmol scale at 60 °C (conventional heating) for 16 h.

^[d]K₂CO₃ was omitted.

^[e]120 °C (MW), 0.2 M.

Significantly, coupling di-tert-butyl diazomalonate with N-methyl phenylalanine amide resulted in a considerable improvement in both yield and diastereoselectivity relative to when diethyl diazomalonate was used (4u versus 4d). When the separate syn and anti diastereomers of 4u were resubjected to the reaction conditions, each equilibrated to a mixture of the diastereomers of 4u (see Mechanistic Studies in the Supporting Information), indicating that the final cyclization step is a reversible process. We therefore evaluated the reaction at a lower temperature for 16 h and also without a base additive, but we did not see improvement in the diastereoselectivity (see Table S5).

Given the high yield and diastereoselectivity of 4u, we chose to further evaluate the coupling of di-tert-butyl diazomalonate with other inputs (Table 3). Piperazinone 4aa was obtained with 89:11 dr from an α-branched aldehyde. Other amino amides also coupled in good yields and high diastereoselectivity as exemplified by 4ab. Finally, the phenylalanine amide provided 4ac, demonstrating that primary amides could be used to prepare piperazinone products, though with more moderate yields and diastereoselectivity.

A variety of different transformations were then employed to further elaborate the piperazinone products 4 (Scheme 2). First, N-acylation of 4a with phenylacetyl chloride afforded 5 in high yield. Decarboxylative saponification of 5 gave solely the anti diastereomer of the mono-carboxylic acid-substituted piperazinone 6 as rigorously determined by X-ray crystallography (see the Supporting Information).^[15] Notably, the free carboxylic acid in 6 provides a versatile site for further diversification. Several N-alkylation reactions were performed on 4a. Reductive amination using formaldehyde provided N,N’-dimethyl piperazinone 7a. Meanwhile, differentially N-substituted piperazinones were prepared by two approaches: reductive amination afforded N-(4-pyridylmethyl) 7b, whereas an S_N2 reaction with benzyl bromide furnished N-benzyl 8. Reduction of 8 with lithium aluminium hydride provided a piperazine 9 that incorporates an α-amino bis-hydroxymethyl tetrasubstituted carbon, which would be difficult to introduce into piperazines by other methods. Finally, we leveraged the pendent ester functionality of 4r to access the fused bicyclic product 10 by cyclization under acidic conditions.

Scheme 2.

Product diversification.

To gain insight into the reaction mechanism, imine 11a was used to evaluate the validity of an imine as a reactive intermediate. When imine 11a was coupled with diazo compound 3a under the optimal reaction conditions for the formation of 4m, we observed a comparable yield of 65% (Scheme 3a). This experiment indicates that the imine is a plausible intermediate in the formation of piperazinones 4. Given this result we reacted imine 11a and [Cp*RhCl₂]₂ and observed formation of rhodacycle 12a in high yield as determined by NMR quantitation relative to an internal standard (Scheme 3b). Moreover, rhodacycle 12a was characterized by X-ray crystallography (see the Supporting Information).^[15] A metallacycle with chelation to both amide and imine functionalities has not been reported, the closest analogy being N,N’-bidentate iridacycles^[12] and rhodacycles^[17] derived from picolinamide substrates.^[18]

Scheme 3.

Mechanistic studies. [a] Yields determined by ¹H NMR integration relative to an internal standard.

Rhodacycle 12a was then tested under our optimal reaction conditions to assess its viability as a reaction intermediate. Amino amide 1b, aldehyde 2c, and diazo compound 3a were coupled in the presence of rhodacycle 12a (10 mol %) and AgSbF₆ (10 mol %), affording 4m in 69% yield (Scheme 3c). Furthermore, coupling diazo compound 3a in the presence of a stoichiometric amount of rhodacycle 12a and AgSbF₆ under otherwise unperturbed reaction conditions provided 4m in 56% yield (Scheme 3d). The results depicted in Schemes 2c-d are consistent with a cationic Cp*Rh(III) rhodacycle as an intermediate along the catalytic pathway to piperazinones 4.

We next investigated whether rhodium chelation is necessary for Cp*Rh(III)-catalyzed N–H functionalization by using amide 13 as a non-chelating N–H bond substrate. When coupled with diazo compound 3a, minimal amounts of N-alkylated product 14 were formed whether or not an aldehyde was present (Scheme 3e and Table S6). Instead, amide 13 was recovered in high yield. These studies show that nitrogen chelation is necessary for amide N–H functionalization. We considered that the amino amide rather than the imine may chelate to rhodium, in which case imine formation could occur after the N–H functionalization step. However, the two-component coupling of amino amide 1b with diazo compound 3a under the three-component reaction conditions provided diketopiperazine 15 in high yield (Scheme 3f), presumably by amine directed Cp*Rh(III)-catalyzed amide N–H functionalization followed by spontaneous cyclization to the diketopiperazine. This result supports imine formation prior to rhodium chelation, thereby enabling imine directed Cp*Rh(III)-catalyzed N–H functionalization. While imines were the first and some of the most extensively used directing groups for C–H functionalization,^[19] to our knowledge this is the first example of an imine directing the functionalization of an N–H bond.

Based on these studies, we propose the following catalytic cycle for the three-component synthesis of piperazinones 4 (Scheme 4). First, amino amide 1 and aldehyde 2 condense to produce imine 11, which then forms nitrogen-chelated rhodacycle 12. Carbene insertion of diazo compound 3 with release of N₂ provides the six-membered rhodacycle 16. Protonolysis could release the putative intermediate 17 to regenerate the Cp*Rh(III) catalyst. Deprotonation alpha to the amide nitrogen in 17, facilitated by the electron withdrawing R⁵ and R⁶ groups, followed by reversible intramolecular addition to the imine would then afford piperazinone 4. Alternatively, direct addition to the imine from 16, which incorporates a C-bound Cp*Rh(III) enolate when either R⁵ and/or R⁶ is a carbonyl group, could be followed by protonolysis to release piperazinone 4 with regeneration of the Cp*Rh(III) catalyst. In relevant studies, a Rh(I)•chiral diene O-bound enolate has been proposed to add to a nitroalkene.^[20]

Scheme 4.

Proposed catalytic cycle.

In conclusion, imines formed in situ from amino amides and aldehydes direct Cp*Rh(III)-catalyzed N–H functionalization with diazo compounds followed by annulation to form piperazinones. A broad range of functionality was compatible with the reaction conditions, and the synthetic versatility of the piperazinones was demonstrated by a variety of product elaborations. The X-ray structural characterization and catalytic competency of a nitrogen-chelated rhodacycle support the proposed imine directed mechanism. The synthesis of additional classes of heterocycles by annulations proceeding via imine directed N–H functionalization are currently underway.