Rhodium-catalysed syn-carboamination of alkenes via a transient directing group.

Alkenes are the most ubiquitous prochiral functional groups--those that can be converted from achiral to chiral in a single step--that are accessible to synthetic chemists. For this reason, difunctionalization reactions of alkenes (whereby two functional groups are added to the same double bond) are particularly important, as they can be used to produce highly complex molecular architectures. Stereoselective oxidation reactions, including dihydroxylation, aminohydroxylation and halogenation, are well established methods for functionalizing alkenes. However, the intermolecular incorporation of both carbon- and nitrogen-based functionalities stereoselectively across an alkene has not been reported. Here we describe the rhodium-catalysed carboamination of alkenes at the same (syn) face of a double bond, initiated by a carbon-hydrogen activation event that uses enoxyphthalimides as the source of both the carbon and the nitrogen functionalities. The reaction methodology allows for the intermolecular, stereospecific formation of one carbon-carbon and one carbon-nitrogen bond across an alkene, which is, to our knowledge, unprecedented. The reaction design involves the in situ generation of a bidentate directing group and the use of a new cyclopentadienyl ligand to control the reactivity of rhodium. The results provide a new way of synthesizing functionalized alkenes, and should lead to the convergent and stereoselective assembly of amine-containing acyclic molecules.

The prominence and importance of nitrogen functionality in biologically relevant molecules is undeniable,[7] and stereoselective methods for introducing nitrogen atoms into organic molecules remain a subject of intense interest. Alkene hydroamination is an emerging technology for the introduction of nitrogen functionality (Fig. 1a).[8,9,10] On the other hand, incorporation of carbon-based coupling partners is more limited, in spite of the crucial role of C-C bond forming reactions in synthesis. Among these, Heck-type approaches are noteworthy for their ability to introduce a carbon fragment in a stereoselective manner under typically mild conditions.[11,12] Both the hydroamination and Heck-type reactions have the same strategic drawback: only one end of the alkene is functionalized. Simultaneous incorporation of both carbon- and nitrogen-based functionalities across an alkene would address this deficiency.

Figure 1

Carboamination Reactions

a, Transition metal catalyzed difunctionalization of alkenes. b, Available carboamination reactions in organic synthesis. c, Rh(III)-catalyzed intermolecular syn-carboamination of alkenes. Ar, aromatic; MCAT, metal based catalyst; Ph, phenyl; Phth, phthalimide.

Established stereoselective carboamination reactions are currently limited and fall into three categories (Fig. 1b). Annulative reactions are well-entrenched and powerful strategies but deliver a cyclic product which inherently limits their impact.[13,14] A handful of intramolecular approaches have been developed wherein one of the reacting partners is tethered to the alkene.[15,16,17] Lastly, there is a growing subset of radical-based reactions which functionalize both ends of the alkene in a carboamination process.[18,19] However, the involvement of radicals in the mechanism means that stereochemistry present in the alkene starting material is typically lost in the course of the reaction. Herein we describe the first stereoselective intermolecular carboamination reaction of alkenes, using enoxyphthalimides as the source of both carbon and nitrogen atoms (Fig. 1c). In the presence of a Rh(III) catalyst, these precursors undergo syn-addition across a variety of disubstituted alkenes in a stereospecific manner, delivering acyclic products containing two contiguous stereocenters in an intermolecular fashion.

We have previously described that enoxyphthalimides undergo Rh(III)-catalyzed reactions with electron-deficient alkenes to deliver cyclopropane adducts (Fig. 2).[20] The proposed mechanism involves the generation of intermediate A, the product of carborhodation of the alkene partner. We hypothesized that the Rh atom is coordinatively unsaturated and thus ligates the enol alkene fragment, which subsequently undergoes migratory insertion to form the C–C bond in the cyclopropane product. Should the Rh atom be coordinatively saturated, intramolecular alkene coordination should be disfavored and we may favor reductive elimination to form the carboamination product. Coordinative saturation of the Rh atom could conceivably occur by intramolecular coordination to a bidentate directing group.

Figure 2

Working hypothesis: Tuning of the directing group to influence reactivity

Ligands on Rh omitted for clarity. Ar, aromatic; L, exogenous nucleophile.

Efforts to install requisite bidentate directing groups[21] on the enoxyamine were frustrated by the instability of the product. We succeeded in overcoming this instability by generating it in situ through the use of a more nucleophilic solvent such as methanol, which we hypothesized would open the phthalimide to form the phthalimide-derived amido ester. Under these conditions, the formation of the carboamination product 3aa is favored over the cyclopropane 4aa in 2.8:1 ratio (Table 1, entry 2). We also observed the formation of 5aa derived from opening of the phthalimide ring. Fortunately, the product 5aa could be converted back to 3aa without erosion of diastereoselectivity by simply heating the crude reaction mixture at 60 °C in toluene after consumption of the starting material 1a (entry 3). Furthermore, it was established that 3aa was formed as a single diastereoisomer, the relative configuration being unambiguously assigned by X-ray crystallography and consistent with a syn-addition process, thereby confirming our initial hypothesis. Selectivity between 3aa and 4aa, however, remained sub-optimal.

Table 1

Optimization of reaction conditions.

entry	Method	Cpx	solvent	ratio 3aa:4aab	Yield 3aa (%)c
1	A	Cp*	trifluoroethanol	1:2.3	30%d
2	A	Cp*	MeOH	2.8:1	49%e
3	B	Cp*	MeOH	3.5:1	60%
4	B	CpiPr	MeOH	-	0%
5	B	Cpt	MeOH	-	<10%
6	B	CpCF3	MeOH	-	<10%
7	B	Cp*Cy	MeOH	8.0:1	69%
8	B	Cp*tBu	MeOH	8.4:1	72%
9	Bf	Cp*tBu	MeOH	14.8:1	82%
10	C	Cp*tBu	MeOH	14.8:1	80% d

Method A: 1a (1 equiv.), 2a (1.2 equiv.), [RhIII] (10 mol %), CsOAc (2 equiv.) in solvent (0.2 M), at rt for 16 h. Method B: 1a (1 equiv.), 2a (1.2 equiv.), [RhIII] (10 mol %), CsOAc (2 equiv.) in solvent (0.2 M), at rt for 16 h then stirred in toluene (0.2 M) at 60 °C for 4 h. Method C: 1a (1 equiv.), 2a (1.2 equiv.), [RhIII] (5 mol %), 1-AdCO2Cs (1 equiv.) in MeOH (0.2 M), at rt for 16 h then stirred in toluene (0.2 M) at 60 °C for 4 h.

Determined by analysis of the unpurified mixture by 1H NMR.

NMR yield.

Isolated yield.

ratio 3aa:5aa:4aa = 2.8:1:1.

1-AdCO2Cs was used as base instead of CsOAc. Ac, acetyl; Ad, adamantyl; Cy, cyclohexyl; Cp, cyclopentadienyl, iPr, isopropyl; Me, methyl; tBu, tert-butyl; Phth, phthalimide; Ph, phenyl; rt, room temperature.

Building on our previous work on cyclopentadienyl ligands, we speculated that control of the chemoselectivity could be achieved through ligand design. Disappointingly, however, when using the monosubtituted CpPr ligand, which performed well in the cyclopropanation reaction, the carboamination product 3aa is not formed (entry 4). Sterically hindered (Cp)[22] or electron-deficient (CpCF3)[23] ligands furnish compound 3aa in poor yields (entries 5 and 6). To our delight, the pentasubstituted Cp*Cy ligand gives the desired product 3aa in 69% yield and good chemoselectivity (3aa:4aa = 8.0:1, entry 7). Further increasing the steric hindrance of the cyclopentadienyl ligand (Cp*Bu) allows the formation of 3aa in an increased yield (72%) with slightly better chemoselectivity (3aa:4aa = 8.4:1, entry 8). Finally, replacing CsOAc with cesium adamantylcarboxylate (1-AdCO2Cs) significantly improves the chemoselectivity (3aa:4aa = 14.8:1) producing the desired product 3aa in 82% yield (entry 9). Importantly, decreasing the catalyst loading from 10 to 5 mol % and using an equimolar amount of base did not affect the efficiency of the reaction (entry 10).

With the optimized reaction conditions in hand, we investigated the generality of the title transformation (Fig. 3a). Structural variations on the N-enoxyphthalimide 1 were examined first (Fig. 3b). The presence of a phenyl ring on substrate 1 proved to be essential. Electron-donating and -withdrawing substituents located at the para, meta, or ortho positions of the phenyl ring are all well tolerated in the reaction, providing the corresponding carboamination adducts 3ba-ka in good to high yields (30–76%). A small handful of products were formed in low yields, which we believe correlates with the relative insolubility of their derived starting materials (specifically N-enoxyphthalimides 1e, 1j and 1k).

Figure 3

Applications of the carboamination reaction

a, General conditions for carboamination of 1,2-disubstituted alkenes. b, Effect of substituents on the N-Enoxyphthalimide. c, Probe of reaction stereospecificity. d, Functionalization of 1,2-disubstituted alkenes. e, Functionalization of mono-substituted alkene. f, Derivatization of the carboamination adduct: Formation of pyrrolidine. Ac, acetyl; Ad, adamantyl; Bn, benzyl; Boc, tert-butoxycarbonyl; Cy, cyclohexyl; Cp, cyclopentadienyl, Et, ethyl; iPr, isopropyl; Me, methyl; Phth, phthalimide; Ph, phenyl; rt, room temperature; tBu, tert-butyl; TBS, tert-butylsilyl.

In order to probe the stereochemical outcome of the reaction, we subjected fumarate and maleate esters to the optimized reaction conditions (2a and 2b, Fig. 3c). The reaction delivered isomeric products 3aa and 3ab in high diastereoselectivity, suggesting that the insertion event is a stereospecific syn addition across the alkene. We next tested a variety of alkenes in our carboamination reaction (Fig. 3d), and were pleased to find that the reaction conditions are mild enough to tolerate sensitive functional groups such as silyl ethers, chloro- and fluoro-alkyls. The corresponding adducts 3ac-ae were isolated in high yields (85–89%) with excellent chemoselectivity. The reaction also proceeds with hindered alkenes leading to 3af and 3ag in excellent yields. Interestingly, in the case of unsymmetrical trans-1,2-disubstituted alkenes, the carboamination reaction takes place with a high control of regioselectivity, leading to products 3ah-aj as the major regioisomers (53–86% yield). In all cases, the most bulky substituent is placed away from the phthalimide group. Additionally, N-phenylmaleimide 2k is a suitable substrate giving the desired product 3ak in 74% yield. Electron-rich alkenes such as 1,2-dihydrofuran 2l and 1,2-dihydropyrrole 2m are also reactive and furnish disubstituted tetrahydrofuran 3al and pyrrolidine 3am in 69% and 25% yield, respectively. Gratifyingly, both heterocycles were obtained as single regio- and diastereoisomers in accordance with a previous report.[24] Finally, by switching to [Cp*CyRh(CH3CN)3](SbF6)2 as catalyst, the scope of the carboamination reaction was expanded to monosubstituted alkenes. Accordingly, when employing ethyl acrylate 2n as coupling partner, the unnatural α-aminoacid derivative 3an is isolated in 53% yield (Fig. 3e).

The carboamination products 3 are versatile entities. Besides their obvious similarity to novel unnatural α-amino acids, they may also be converted into pyrrolidines 7 (Fig. 3f). Deprotection of the phthalimide group followed by cyclization affords the 1,2-dihydropyrrole 6 (dr = 10:1, 85% yield). The latter can be reduced under heterogeneous conditions to yield pyrrolidine 7 in high diastereoselectivity (dr >20:1).

In order to interrogate our mechanistic hypothesis, we probed whether the delivery of the phthalimide moiety occurs via an intra- or intermolecular process. To this end, we performed a crossover experiment by submitting an equimolar mixture of N-enoxyphthalimides 1f and 1l to our optimized reaction conditions (Fig. 4a). No crossover adduct 8 is formed suggesting that, in accordance with our initial proposal (Fig. 2b), delivery of the phthalimide moiety takes place in an intramolecular fashion. Moreover, when subjecting product 3aa back to the reaction, the formation of 5aa was not observed (Fig. 4b). This result suggests that the opening of the phthalimide group occurs prior to formation of the final product 3aa. Thus, the adduct 5aa might be formed first and cyclizes back in the course of the reaction to give 3aa. This assumption was further confirmed by monitoring the reaction progress by 1H NMR (see Supporting Information for details). To further elaborate on this idea, we investigated the reactivity of a bidentate substrate. Attempts to open the phthalimide group with MeOH were unsuccessful due to the instability of the product. However, the parent substrate 1m proves to be relatively more stable and was subjected in the carboamination reaction (Fig. 4c). The expected product 3ma is indeed formed, albeit in moderate yield (35%). A control experiment demonstrates that the carboamination product 3ma does not open in presence of exogeneous pyrrolidine under our standard reaction conditions (see Supporting Information for details). Taken together, these results support our hypothesis that the directing group might be bidentate and emerge from in situ opening of the phthalimide moiety.

Figure 4

Reaction mechanism: study and proposal

a, Crossover experiment. b, Probe of the formation of 5aa. c, Reactivity of a secondary amide in the carboamination reaction. d, Proposed mechanism for the carboamination reaction. Ligands on Rh and phthalimide substituents omitted for clarity. Ad, adamantyl; Cp, cyclopentadienyl, Me, methyl; Ph, phenyl; Phth, phthalimide; tBu, tert-butyl; rt, room temperature.

Based on these mechanistic experiments, we propose the following catalytic cycle (Fig. 4d). First, in presence of MeOH and a base, the N-enoxyphthalimide 1a can reversibly open to form intermediate II (route a, Fig. 4d). The active Rh(III) catalyst then undergoes an irreversible C-H activation at the vinylic position leading to the five-membered rhodacycle III. Alternatively, we cannot rule out the possibility that the C-H activation event precedes the opening of the phthalimide group (IV to III, route b). Subsequent migratory insertion of alkene 2 generates the coordinatively saturated Rh(III) complex V with coordination of the ester group to the metal. We postulate that the bidentate directing group formed in situ stabilizes intermediate V inhibiting both competitive migratory insertion into the enol alkene and β-H-elimination,[25,26] and instead favoring reductive elimination. An oxidative addition of the N-O bond into Rh(I) followed by protonation/tautomerization of the enol liberates the opened product 5a with concomitant regeneration of the active Rh(III) catalyst. Finally, in the course of the reaction, the phthalimide group is reformed to afford product 3a. Contemplating the proposed mechanism, the origin of the chemoselectivity could be rationalized by the solvent effect (entry 1 vs 2). When using MeOH as solvent the initial opening of the phthalimide moiety prevails favoring the formation of intermediate III and therefore the carboamination pathway. Conversely, less nucleophilic trifluoroethanol tends to preserve the integrity of the phthalimide, and thus the cyclopropanation pathway is preferred.

In conclusion, we have developed a syn-carboamination of disubstituted alkenes. The reaction utilizes enoxyphthalimides and a Rh(III) catalyst. Ligand development has revealed a new bulky cyclopentadienyl group that alters the inherent chemoselectivity of a reaction. The use of methanol as a solvent is key as was the observation that the phthalimide group undergoes in situ ring opening. Mechanistic experiments suggest that the basicity of the pendant carbonyl stabilizes a Rh(III) intermediate by coordinative saturation which leads to reductive elimination rather than cyclopropanation. Efforts are underway to broaden this reaction and develop an asymmetric version of the transformation.