Synthesis of Enantiomerically Enriched Imidazolidin-2-Ones via Asymmetric Pd-Catalyzed Alkene Carboamination Reactions

Abstract

A catalyst composed of Pd₂(dba)₃ and (S)-Siphos-PE is effective for the enantioselective coupling of N-allyureas with aryl bromides to afford 4-substituted imidazolidin-2-ones with up to 95% ee. Added water leads to significantly improved enantioselectivities with electron-poor aryl halide substrates. The mechanism of these transformations is discussed, and evidence is presented which suggests that, in contrast to other asymmetric alkene carboaminations, C–C bond-forming reductive elimination is the enantiodetermining step of these reactions.

Chiral 4-substituted imidazolidin-2-ones are displayed in a myriad of biologically active compounds, including potent HIV protease inhibitors.^[i] However, despite their significance, the preparation of these compounds via asymmetric catalysis has rarely been reported.^[ii] In addition, no existing catalytic asymmetric reaction allows for the generation of a C–C bond during the course of imidazolidin-2-one formation. Most other routes to enantiopure imidazolidin-2-ones involve the synthesis of chiral 1,2-diamines, which often requires several synthetic steps and can be particularly lengthy when the two amino groups bear different substituents.^[iii,iv]

In this communication we describe the first catalytic asymmetric alkene carboamination reactions between N-allylurea derivatives and aryl or alkenyl halides.^[v,vi] These reactions provide a simple route to enantiomerically enriched 4-(arylmethyl)imidazolidin-2-ones, and these studies illustrate the utility of water as an additive in an enantioselective Pd-catalyzed alkene difunctionalization reaction. Moreover, the results of these experiments suggest that C–C bond-forming reductive elimination may be the enantiodetermining step in transformations of N-allylurea substrates. This contrasts with other asymmetric metal-catalyzed alkene carboamination reactions, in which the absolute stereochemistry appears to be set during alkene aminometallation.^[vii]

(1)

Our prior studies on Pd-catalyzed asymmetric carboaminations of N-Boc-pent-4-enylamine demonstrated that use of (R)-Siphos-PE^[viii] as a ligand provides 2-benzylpyrrolidine derivatives with good levels of enantioselectivity.^[viia] Thus in our efforts to generate enantiomerically enriched imidazolidin-2-ones we initially examined the use of a catalyst composed of Pd₂(dba)₃ and (R)-Siphos-PE for the asymmetric conversion of 1b to 2b. Unfortunately, this catalyst provided 2b in only modest enantioselectivity [Eq. (1)].^[ix] However, we were gratified to find that the diastereomeric ligand (S)-Siphos-PE provided 2b in 93% yield and 79% ee. Despite this promising result, a survey of other phosphine and phosphoramidite ligands did not lead to further improvements in asymmetric induction.

In efforts to both increase the asymmetric induction in Pd/(S)-Siphos-PE-catalyzed carboamination reactions of N-allyl ureas, and to explore the effect of nitrogen nucleophilicity on asymmetric induction, we prepared a series of substrates bearing different aromatic groups on the cyclizing nitrogen atom. As shown in Table 1, the level of asymmetric induction increased with the increasing electron-withdrawing ability of the p-substituent on the N-aryl moiety of 1. Unfortunately, the chemical yields also decreased due to the diminished reactivity of these substrates, and unreacted starting material was observed in crude reaction mixtures. To address this problem the reaction temperature was increased to 115–120 °C, and the desired products were generated in good chemical yield (entries 6–7) without a significant decrease in asymmetric induction.^{[ x ]} The best result was obtained with p-nitrophenyl substituted substrate 1f, which was converted to 2f in 81% yield and 92% ee.

Table 1.

Electronic effects.^[a]

Entry	X	Yield (%)[b]	ee (%)
1	N(Bn)Me (1a)	70 (2a)	73
2	OMe (1b)	93 (2b)	79
3	H (1c)	90 (2c)	78
4	Br (1d)	45 (2d)	83
5	CN (1e)	46 (2e)	89
6[c]	CN (1e)	87 (2e)	86
7[c]	NO2 (1f)	81 (2f)	92

^[a]Conditions: 1.0 equiv 1, 2.0 equiv p-^tBu-C₆H₄-Br, 2.0 equiv NaO^tBu, 2 mol % Pd₂(dba)₃, 6 mol % (S)-Siphos-PE Toluene (0.2 M), 90 °C, 12 h.

^[b]Isolated yield (average of two or more runs).

^[c]The reaction was conducted at 120 °C in xylenes.

In order to explore the scope of the asymmetric carboamination reactions, substrate 1f was coupled with a range of different aryl halide electrophiles. We initially encountered difficulty obtaining reproducible enantioselectivities in reactions of electron-deficient aryl bromides with 1f. After some exploration, we discovered that addition of 2.0 equiv of water to the reactions led to significantly improved and reproducible enantioselectivities.^[xi] The coupling of 2-bromonaphthalene or 3-bromoanisole with 1f was not significantly influenced by the addition of water (entries 11 and 14). Addition of water had a negative impact on the reaction of 1f with 4-bromoanisole, but addition of 40 mol % of trifluoroacetic acid provided results superior to those obtained with either added water or no additive (entry 13). Efforts to employ alkenyl halides as electrophiles in the asymmetric carboamination of 1f were unsuccessful. However, the reaction of substrate 1e with 2-bromovinyltrimethylsilane proceeded in 65% yield and 86% ee (entry 15). The coupling of 1e with aryl bromides generated cyclic urea products with lower enantioselectivies than were obtained in analogous reactions of 1f (entries 16–20). Use of aryl iodides as electrophiles in reactions of 1e and 1f led to diminished enantioselectivities as compared to reactions of the corresponding aryl bromides (entries 8–9 and 18–19). However, addition of water to the reaction of 4-iodobenzotrifluoride led to a slightly higher ee than was obtained with the aryl bromide under similar conditions (entries 16–17).

The reactivity of substrates bearing disubstituted alkenes was also briefly explored. The 2-methylallyl urea derivative 3 was converted to cyclic urea 4 in 72% yield and 76% ee although a reaction temperature of 135 °C was required to effect complete conversion [Eq. (2)]. Addition of water or TFA did not have a positive effect on either the chemical yield or the enantioselectivity of this transformation. The 3-phenylallyl urea substrate 5 was unreactive at 135 °C, and only trace amounts of product were obtained at 160 °C [Eq. (3)]. In order to determine the stereochemistry of the 1,2-addition reaction, Z-deuterioalkene substrate (Z)-d-1f was subjected to the optimized reaction conditions. This transformation proceeded with syn-addition to the alkene to afford d-2f in 85% yield, 91% ee, and 7:1 dr [Eq. (4)].^[xii]

(2)

(3)

(4)

To illustrate the feasibility of deprotecting the N-aryl urea products we examined removal of the p-nitrophenyl group from 2m. In this representative case, cleavage of the N-aryl substituent was cleanly accomplished using a three-step procedure that required only a single chromatographic purification. As shown below, hydrogenation of 2m followed by treatment with acetic anhydride and then oxidation with ceric ammonium nitrate provided 6 in 88% overall yield [Eq.(5)].

(5)

Several of the observations described above are quite surprising, particularly when results obtained in the Pd/Siphos-PE-catalyzed asymmetric carboaminations of ureas are compared to those for related asymmetric Pd-catalyzed carboamination reactions of N-boc-pent-4-enylamines (Scheme 1). For example, the Pd₂(dba)₃/(S)-Siphos-PE catalyst provides (S)-imidazolidin-2-one products 2 in reactions of 1e–f, whereas the same catalyst system afforded pyrrolidine enantiomer 8 from 7.^[xiii] Thus, despite the similarities of the two substrates, the two reactions proceed with the opposite absolute asymmetric induction. The positive effect of water on asymmetric induction in Pd-catalyzed reactions has rarely been observed,^[xiv] and the water effect is considerably different for the imidazolidin-2-one vs. pyrrolidine systems. Addition of water in reactions of 1e or 1f with electron-deficient aryl bromides led to significantly improved asymmetric induction. In contrast, addition of water to reactions of N-boc-pent-4-enylamine derivatives had a highly deleterious effect, as no desired product was obtained in these transformations.^[xv]

Scheme 1.

Stereochemistry of urea vs. pyrrolidine formation.

Despite the differences in the absolute product stereochemistry in reactions of 1 vs. 7, these transformations appear to proceed through similar mechanisms and intermediates. In both cases syn-1,2-addition to the alkene is observed, and the ratio of ligand to Pd does not influence asymmetric induction in either case.^[viia] The mechanism of both reactions proceeds via oxidative addition of the aryl halide to Pd(0) followed by substrate deprotonation and substitution to provide alkene-bound palladium(aryl)(amido) complexes 9a–b and 14a–b (Scheme 2).^[vi] These complexes undergo syn-1,2-migratory insertion into the Pd–N bond to yield 10 and 16,^[xvi] which then undergo reductive elimination to provide the observed major products 2 and 8.

Scheme 2.

Mechanism of urea formation vs. pyrrolidine formation.

The differences in absolute stereochemistry of products obtained in reactions of the urea substrates vs. the N-Boc pentenylamine derivatives are likely due to a change in the enantiodetermining step of the catalytic cycle. The product stereocenter is generated during the migratory insertion step, but two steps could potentially be enantiodetermining in this catalytic cycle: (a) the migratory insertion step (e.g., 14b to 16); or (b) the subsequent reductive elimination step if migratory insertion is reversible (e.g., 10 to 2).^[xvie] In our prior studies^[viia] we suggested that with the relatively electron-rich N-boc pentenylamine substrate 7^[xvii] the aminopalladation is likely irreversible,^[xviii] and relatively fast reductive elimination from 16^[xix] would provide the observed major pyrrolidine stereoisomer 8 following enantiodetermining aminopalladation (Scheme 2, bottom).

In contrast, the influence of N-aryl group electronic properties and anionic ligands on enantioselectivity in reactions of N-allylureas are consistent with a mechanism for imidazolidin-2-one formation that proceeds via reversible aminopalladation followed by enantiodetermining reductive elimination (Scheme 2, top). Electron-withdrawing N-substituents on the cyclizing nitrogen atom have been shown to decrease the rate of C–C bond-forming reductive elimination from complexes related to 10–11 and 15–16,^[xvia,c] and to promote β-amidate elimination (retro-aminopalladation) from these complexes.^[xvie] Thus, equilibration of 11 × 9b × 9a × 10 is likely more facile for electron-poor substrates 1e–f. Reductive elimination then occurs more rapidly from 10 rather than 11 due to the chirality of the ligand to yield the observed major enantiomer 2. In contrast, equilibration (retro-aminopalladation) of 11 and 9b may occur less readily for more electron-rich derivatives 1a–c, thus a greater amount of the minor enantiomer (i.e. lower enantioselectivity) is obtained with these substrates due to an inherent preference for insertion via 9b (as is the case for 14b in the N-Boc series).

The observed influence of anionic ligands^[xx] (iodide from aryl iodides and hydroxide from added water) on these transformations may be due to their effect on the rate of the C–C bond-forming reductive elimination step, which could perturb the equilibration between 11 and 9b.^{[ xxi ]} Stereoselectivies are improved for the combination of small hydroxide ligand and electron-poor aryl halides, which is consistent with the electron-rich hydroxide ligand slowing reductive elimination (and thereby facilitating equilibration) for these types of substrates that are otherwise prone to undergo relatively rapid C–C bond formation.^[xxi] This effect is minimized for electron-neutral or -rich aryl halides where reductive elimination is slow relative to electron-poor aryl halides. In contrast, the relatively large iodide ligand may increase the rate of reductive elimination through a steric effect,^[xxii] which is most pronounced with the electron-neutral or -rich aryl halides. This would be expected to disfavor equilibration of 11 and 9b and lead to diminished enantioselectivities as observed. The positive effect of trifluoroacetic acid on the reaction of 4-bromoanisole with 1f is also presumably due to coordination of a trifluoroacetate anion to 10 and 11, although the precise nature of this effect is not clear. The pronounced effect of anionic ligands on the enantioselectivity of reactions of 1e–f likely does not arise via anion binding to the metal prior to or during the aminopalladation step. The insertion of alkenes into Pd–N bonds has previously been shown to proceed via 4-coordinate alkene-bound complexes such as 9a–b or 14a–b, and insertion from 5-coordinate species appears to be unfavorable.^[xvi]

In conclusion, we have developed a new catalytic asymmetric synthesis of 4-benzyl-imidazolidin-2-one derivatives via enantioselective carboamination reactions. The N-allylurea substrates are readily available (one step from commercially available materials), and products are generated in good yield and up to 95% ee. Importantly, these studies illustrate that the enantiodetermining step in asymmetric Pd-catalyzed carboamination reactions may be influenced by substrate structure, and that substrate electronics and anionic additives also greatly affect levels of asymmetric induction. These observations will likely be of significant utility in the future development of other enantioselective alkene difunctionalization reactions that involve potentially reversible alkene insertion processes.

Table 2.

Scope of Pd-catalyzed asymmetric carboamination.^[a]

Entry	G	R–X	Additive[b]	Product	Yield (%)[c]	ee (%)
1	NO2	p-CF3-C6H4–Br	–	2g	56	77
2	NO2	p-CF3-C6H4–Br	H2O	2g	72	95
3	NO2	m-CF3-C6H4–Br	H2O	2h	68	88
4	NO2	o-CF3-C6H4–Br	H2O	2i	56	83
5	NO2	p-PhC(O)-C6H4–Br	H2O	2j	82	86
6	NO2	p-F-C6H4–Br	H2O	2k	65	94
7	NO2	p-Cl-C6H4–Br	H2O	2l	74	93
8	NO2	p-tBu-C6H4–Br	–	2f	81	92
9	NO2	p-tBu-C6H4–I	–	2f	60[d]	47
10	NO2	Ph–Br	–	2m	83	89
11	NO2		–	2n	71	89
12	NO2		H2O	2o	80	87
13	NO2	p-MeO-C6H4–Br	TFA	2p	80	90
14	NO2	m-MeO-C6H4–Br	–	2q	75	83
15	CN		–	2r	65	86
16	CN	p-CF3-C6H4–Br	H2O	2s	58	77
17	CN	p-CF3-C6H4–I	H2O	2s	77	80
18	CN	p-Me-C6H4–Br	–	2t	86	85
19	CN	p-Me-C6H4–I	–	2t	89	73
20	CN	p-MeO-C6H4–Br	–	2u	73	84

^[a]Conditions: 1.0 equiv 1e or 1f, 2.0 equiv ArX, 2.0 equiv NaO^tBu, 2 mol % Pd₂(dba)₃, 6 mol % (S)-Siphos-PE, xylenes (0.2 M), 115 °C (substrate 1f) or 120 °C (substrate 1e), 14–18 h.

^[b]2.0 equiv water or 40 mol % TFA were added to the reaction mixture prior to heating.

^[c]Isolated yield (average of two or more runs).

^[d]This material contained ca. 35% of an unidentified regioisomer.