Marie
Peng
a,
Denis
Ari
a,
Thierry
Roisnel
a,
Henri
Doucet
a and
Jean-François
Soulé
*b
aUniv. Rennes, CNRS UMR6226, Rennes F-3500, France
bChimie ParisTech, PSL University, CNRS, Institute of Chemistry for Life and Health Sciences, 75005 Paris, France. E-mail: jean-francois.soule@chimieparistech.psl.eu
First published on 1st August 2023
We introduce a versatile Rh(I)-catalyzed cascade reaction, combining C(sp2)–H bond functionalization and amidation between N-arylphosphanamines and acrylates. This innovative approach enables the rapid synthesis of dihydroquinolinone scaffolds, a common heterocycle found in various pharmaceuticals. Notably, the presence of the phosphorus atom facilitates the aniline ortho-C(sp2)–H bond activation prior to N–P bond hydrolysis, streamlining one-pot intramolecular amidation. Moreover, we demonstrate the applicability of this reaction by synthesizing an antipsychotic drug. Detailed mechanistic investigations revealed the involvement of a Rh–H intermediate, with substrate inhibition through catalyst saturation.
The second issue to face is the selectivity between alkenylation versus alkylation. For instance, in 2015, Liu and co-workers demonstrated that the installation of a transient acetamido as a directing group on aniline partners allows the synthesis of 2-quinolinones through a cascade reaction of Pd-catalyzed C(sp2)–H alkenylation – cyclization reaction (Fig. 1C(2)).6 Therefore, a subsequent hydrogenation step of the double bond is required to deliver dihydroquinolinones.7 The alkenylation versus alkylation selectivity generally depends on the choice of the directing group. Indeed, Chang and co-workers demonstrated that the acetamido group affords alkenylation, while pyridine, pyrazole, and pyrimidine lead to alkylation products using [Cp*IrCl2] catalyst (Fig. 1C(3)).8 Nevertheless, utilizing Rh(III)-catalysis, the non-removable directing group of pyridine enables alkenylation even within the annulation process.9 Alternatively, allylic alcohols can be used in Rh(III)-catalyzed ortho-directed C–H bond alkylations of anilines using pyrimidine as non-removable directing group.10 Following the discovery that trivalent phosphorus [P(III)] acts as directing group for the C(sp2)–H bond functionalization of phosphines11 or indoles,12 and given the fact that P(III) favors alkylation products with acrylates using Rh(I) catalysts,13 we hypothesized that if N-arylphosphanamines could be alkylated at ortho-position with acrylates, such intermediate would rapidly undergo intramolecular amidation reaction driven by N–P bond hydrolysis provided that suitable conditions are found (Fig. 1C(4)). Herein, we describe the reaction development with mechanistic aspects showcasing that trivalent phosphorus acts as a traceless directing group in Rh(I)-catalyzed cascade reaction of C(sp2)–H bond activation/alkylation – amidation of anilines with acrylate derivatives for the one-pot synthesis of dihydroquinolinones.
Entry | L | Additive | Solvent | 3a (%) |
---|---|---|---|---|
a Determined by GC-analysis using n-dodecane as internal standard, isolated yield is shown in parentheses. b Yield obtained after HCl hydrolysis. c Yield of 4a. d Methyl acrylate instead of tert-butyl acrylate. e Ethyl acrylate instead of tert-butyl acrylate. | ||||
1 | — | — | Toluene | 42b |
2 | — | — | 1,4-Dioxane | 12b |
3 | — | — | ClCH2CH2Cl | 24 |
4 | — | — | DMF | 40 |
5 | — | NaOAc (0.3 equiv.) | DMF | 30 |
6 | — | AlMe3 (0.2 equiv.) | DMF | 20 |
7 | L1 | — | DMF | 54 |
8 | L2 | — | DMF | 36 |
9 | L3 | — | DMF | 46 |
10 | L4 | — | DMF | 45 |
11 | L5 | — | DMF | 22 |
12 | L6 | — | DMF | 58 |
13 | L6 | MS 4 Å (50 mg) | DMF | 21 (51)c |
14 | L6 | H2O (2 equiv.) | DMF | 62 |
15 | L6 | H2O (6 equiv.) | DMF | 75 |
16 | L6 | H2O (10 equiv.) | DMF | 90 (81) |
17 | L6 | H2O (20 equiv.) | DMF | 54 |
18 | — | H2O (10 equiv.) | DMF | 35 |
19d | L6 | H2O (10 equiv.) | DMF | 85 |
20e | L6 | H2O (10 equiv.) | DMF | 84 |
With the aforementioned optimal reaction conditions established, the substrate scopes of this reaction were then examined (Scheme 1). First, different substituents on the aromatic ring of the aniline moiety were examined. Electron-donating (MeO, Me) and electron-withdrawing (OCF3, F, Cl) groups at the para-position of the N-atom were tolerated providing dihydroquinolinones 3b–f in good to high yields. However, the conditions were incompatible with Br substituent on the N-arylphosphanamine partner. When meta-substituted phosphanamines were employed, the C(sp2)–H bond alkylation regioselectivity occurred at the less hindered position, whatever the substituent, as exemplified by products 3h–k containing F, Cl, Me, or MeO groups. From N-(3,5-difluorophenyl)-1,1-diisopropyl-N-methylphosphanamine, dihydroquinolinone 3l was isolated in 72% yield. Reaction with ortho-fluoro-N-arylphosphanamine 1m gave 3m in a moderate yield of 42%. Other alkyl groups on the N-atom such as Et or n-Bu were also tolerated affording 3n–p in 75–83% yields.15 Beyond N-alkyl phosphanamines, 1,2,3,4-tetrahydroquinoline was also subjected to this C(sp2)–H bond annulation to afford tricyclic amide 3q in 70% yield. α-Substituted acrylates such as dimethyl 2-methylenesuccinate (2b) and methyl methacrylate (2c) were also successfully coupled with 1a to produce 3-substituted N-methyl-dihydroquinolinones 3r and 3s in 57% and 60% yield, respectively.
Satisfied by these results, we then moved to extend this trivalent phosphorus traceless directing group strategy to ortho-C(sp2)–H bond functionalization of anilines with styrene derivatives (Scheme 2). Accordingly, ortho-alkylated N-methyl aniline 5a was obtained in 60% yield from 1a and styrene (2d) using the same catalytic system, namely 2 mol% [RhCl(COD)]2 associated with 4 mol% of L6 in the presence of 10 equivalents of water in DMF. Noteworthy, the C–P bond was not fully hydrolyzed when the reaction was performed without water. Electron-rich 1-(tert-butyl)-4-vinylbenzene displayed a higher reactivity than styrene, as exemplified by forming 5b and 5c in 65% and 73% yield, respectively. This reactivity trend was also confirmed with 1-methoxy-4-vinylbenzene and 2,4-dimethyl-1-vinylbenzene, with which 1a was coupled to give the ortho-alkylated N-methyl anilines 5d and 5e in 75% and 69% yield, respectively. Conversely, the reaction was more sluggish with styrenes holding an electron-withdrawing group such as F and Cl, as the resulting products 5f and 5g were isolated in only 32% and 42% yields.
Having established a method to construct dihydroquinolinones with good functional group compatibility, we investigated the feasibility of applying it to prepare aripiprazole N-methylated analog 8. Aripiprazole treats schizophrenia, bipolar disorder, major depressive and tic disorders.16 In 2020, the aripiprazole market was dominated by two brand companies (Otsuka & Bristol-Myers Squibb) and is expected to grow at a CAGR of 4.60% in the forecast period of 2022–2029. The majority of previous routes toward aripiprazole (or its N-alkyl congeners) involved multi-step synthesis for the construction of the dihydroquinolinone scaffold17 (e.g., intramolecular Friedel–Craft reaction followed by Pd-catalyzed hydrogenation of double bond,18 or through a Beckmann rearrangement from indanone derivative). Owing to the increasing demand for aripiprazole and its N-alkyl analogs, and the lack of synthetic routes for the construction of dihydroquinolinones from simple starting materials prompted us to develop an efficient route to aripiprazole N-methylated analog 8, employing our Rh(I)-catalyzed cascade reaction of C(sp2)–H bond alkylation – amidation of anilines (Scheme 3). A gram-scale reaction (5 mmol) performed from 1,1-diisopropyl-N-(3-methoxyphenyl)-N-methylphosphanamine with 2a gave 0.54 g of 3j (57% yield). Then, classical deprotection of phenol group by BBr3, followed by Williamson reaction using 2 equivalents of 1,4-dibromobutane in the presence of K2CO3 afforded 7 in good yield. Final coupling with 1-(2,3-dichlorophenyl)piperazine in the presence of KI as a relay-nucleofuge and triethylamine as base led to the formation of desired aripiprazole N-methylated analog 8 in 70% yield.
Intrigued by the exact role and the becoming of the phosphorus group along this reaction, we started a mechanistic investigation by carrying out a set of control experiments. No reaction occurred from P,P-diisopropyl-N-methyl-N-phenylphosphinic amide, indicating that trivalent phosphorus is not oxidized at least before the C(sp2)–H bond cleavage. The use of classical acetamide as a directing group or reaction from N-methylaniline also failed to deliver cyclized amide 3a (Fig. 2A). These results indicate that trivalent phosphorus is the active directing group. A standard reaction between phosphanamine 1a and 2a in the presence of 1 equivalent of 3-fluoro-N-methylaniline revealed that there is no phosphorus-directing group scrambling (Fig. 2B). A careful analysis of the reaction outcomes by GC-MS and 31P NMR has allowed us to identify the formation hydroxydiisopropylphosphane (9a) as the side product,19 which may explain that -P(i-Pr)2 acts as a traceless directing group rather than a transient directing group (Fig. 2C).20 To completely rule out the intermediacy of the NH amide substrate, we conducted a control experiment in the presence of one equivalent of 9a, which did not yield the product 3a (Fig. 2D). To determine the role of L6, we decided to prepare the well-defined Rh(L6)(COD) complex, but it was completely inactive for this cascade reaction (Fig. 2E). This result may suggest that L6 has an indirect role, possibly in a catalyst regeneration process, by facilitating the decomplexation of 9a through an unknown mechanism. To get more information on the structure of rhodium intermediates, we then conducted a stoichiometric reaction between [RhCl(COD)]2 and 1a with in situ following by NMR spectroscopies (Fig. 2F). At room temperature, a mixture of two distinct phosphorus containing Rh complexes was observed within two hours. It is generally accepted that treatment of complexes [{Rh-(COD)(μ2-Cl)}2] with phosphine ligands in the absence of silver salts smoothly affords neutral μ2-bridged dinuclear rhodium complexes of the type [{Rh(phosphine)2(μ2-Cl)}2].21 The signal at 87.1 ppm (d, 1JRh,P = 156.7 Hz) is attributed to [{Rh(1a)2(μ2-Cl)}2]; while after isolation, we can attribute the signal at 30.4 ppm (d, 1JRh,P = 140.5 Hz) to [{Rh(9a)2(μ2-Cl)}2]. This complex might be formed through the in situ hydrolysis of P–N bond by water. After 24 h, the proportion of [{Rh(9a)2(μ2-Cl)}2] increases, indicating that it is the resting state catalyst. Although multiple doublet signals are also observed by 31P NMR data, a characteristic Rh–H signal appears as a doublet of doublets at −21.66 ppm (1JRh,H = 34.1 Hz, 2JP,H = 27.6 Hz) in 1H NMR, proving that a C–H bond activation mechanism rather than a CMD mechanism (Fig. 2F). To gain further insight into the reaction mechanism, deuterated labelling experiments were carried out in the standard reaction conditions (Fig. 2G). A standard reaction performed using 10 equivalents of deuterated water gave 57% deuterium incorporation at ortho-position of the NMeP(i-Pr)2 group. This deuterium labelling experiment suggests that the C(sp2)–H bond cleavage is reversible. Moreover, 69% and 20% deuterium incorporations were detected at the α and β-positions of the ester group of 3d. In addition, the deuterium incorporation in the NMe group may result from the formation of amide rhodacycle through C(sp3)–H bond activation. The control experiment involving 3d under deuterium labelling conditions does not show any deuterium incorporation in 3d. The observed H/D isotopic exchanges likely occur due to the formation of Rh–D species, similar to a process described by Martin and Milstein in the C–H bond activation of benzylphosphine by Rh(I) complexes.22 Furthermore, the incorporation of deuterium at the α-position of the ester group should result from an insertion/β-H elimination process.
Then, we conducted a kinetic study to obtain a better picture of the reaction pathway (Fig. 3). First, we plotted the kinetic profile of the reaction by following the consumption of 1a, the production and consumption of 4a by 31P NMR, while the production of 3a was followed by 1H NMR (Fig. 3A). The Gaussian-shape kinetic profile of 4a indicates that k1 > k2 meaning that 4a is a latent intermediate of the reaction. However, we have never observed hydrolyzed aniline 10a during the reaction, possibly due to a spontaneous cyclization affording 3a. Some de facto approximations were formulated. If [10a] ≈ 0 during the reaction, then k3 ≫ k2; steps 2 and 3 can be considered as a single operation, and the rate determining step (RDS) is the hydrolysis of N–P(i-Pr)2 bond. Our attention next turned toward determining the factors influencing the kinetic efficiency of Rh(I)-catalyzed cascade C(sp2)–H bond alkylation – amidation of anilines. Reaction progress kinetic analysis revealed that catalyst deactivation was not caused by substrates or product formation, including 9a.17 Then, we measured the initial reaction rate for the model reaction as a function of the concentration of water, catalyst, tert-butylacrylate (2a) and phosphanamine 1a in DMF at 160 °C. Interestingly, we observed a zero-order dependence on water for the consumption of 1a, and a first-order dependence for the production of 3a (Fig. 3B). These observations indicate that the water is required for the second step, namely the hydrolysis of N–P(i-Pr)2 bond, and its presence (up to 10 equivalents) has a negligible influence on the Rh(I) catalytic activity.23 In contrast, first-order dependence on Rh/L6 catalytic system was observed for the consumption of 1a (Fig. 3C). The first-order dependence of the reaction rate on acrylate concentration was obtained from the linear plot of initial rate (Δ[1a]/Δt) vs. initial acrylate concentration ([2a]) (Fig. 3D). Saturation kinetics were observed with respect to phosphanamine concentration ([1a]) when similar experiments were carried out by measuring the initial rate over a range of 1a concentrations (Fig. 3E). This rate inhibition by phosphanamine 1a might have been anticipated due the strong ability of P(III)-ligands to coordinate Rh delivering a fully-coordinate stable complexes as resting state catalysts.
![]() | ||
Fig. 3 Kinetic study: (A) reaction profile, (B) determination of H2O order, (C) determination of catalyst order, (D) determination of 2a order, (E) determination of 1a order. |
Based on our mechanistic and kinetic investigations and the DFT calculation by Bai, Lan on Rh(I)-catalyzed C7-alkylation of indole N–Pt-Bu224 – which demonstrated that μ2-bridged dinuclear rhodium complexes preferentially act as active catalysts because of the endothermic monomerization process – we propose the catalytic cycle depicted in Fig. 4. [RhCl(COD)]2 complex reacts with phosphanamine 1a to form the active catalytic species I. Then, oxidative addition occurs to deliver the Rh(III)–H hydride II. Acrylate 2a may followed by 1,2-insertion to yield III. Then, reductive elimination affords IV. In the presence of water, the coordinated functionalized phosphanamine is prompted to hydrolysis to give the desired product 3a along with Rh complex V surrounded by hydroxydiisopropylphosphane (9a). The next catalytic cycle starts by a ligand exchange between 9a and 1a to give I. However, V or I may react with phosphine derivatives 9a or 1a in an inhibitory manner to form nonproductive coordination complexes [{Rh(9a)2(μ2-Cl)}2] and [{Rh(1a)2(μ2-Cl)}2]. At this stage, the exact role of L6 remains unclear but it may help the decomplexation of one phosphine to regenerate the active catalysts.
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2252526. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3sc02992a |
This journal is © The Royal Society of Chemistry 2023 |