Rh(III)-Catalyzed C–H Activation-Initiated Directed Cyclopropanation of Allylic Alcohols

Abstract

We have developed a Rh(III)-catalyzed diastereoselective [2+1] annulation onto allylic alcohols initiated by alkenyl C–H activation of N-enoxyphthalimides to furnish substituted cyclopropyl-ketones. Notably, the traceless oxyphthalimide handle serves three functions: directing C–H activation, oxidation of Rh(III), and, collectively with the allylic alcohol, in directing cyclopropanation to control diastereoselectivity. Allylic alcohols are shown to be highly reactive olefin coupling partners leading to a directed diastereoselective cyclopropanation reaction, providing products not accessible by other routes.

Graphical Abstract

Biological and synthetic targets containing cyclopropane units have intrigued organic chemists for years as a result of their unique properties and the synthetic challenges.¹ A number of powerful methods have been developed for the stereoselective synthesis of cyclopropane motifs.² These methods largely share a common approach of an alkene that undergoes a [2+1] annulation with carbenes, metal carbenes, or metal-carbenoid species. In particular, allylic alcohols have been exploited as coupling partners in cyclopropanation reactions for their leverageable, pendent hydroxyl group. Ultimately, this handle provides regio- and diastereoselective cyclopropanations.

Two methods have emerged as preferred techniques for the cyclopropanation of alkenes: Simmons-Smith type reactions and catalyzed diazo decompositions. The Simmons-Smith approach features stoichiometric zinc reagents to aid both the formation and transfer of carbenoid species from simple methylene sources. Similarly, metal-catalyzed diazo decomposition is a broadly powerful reactivity manifold for the cyclopropanation of alkenes, with Rh,³ Ru,⁴ Pd,⁵ Cu,⁶ Co,⁷ and Fe⁸ catalysts utilized for their carbenoid formation and transfer capabilities. Notably, both modes of reactivity have also been rendered asymmetric when using prochiral alkenes.⁹

With regards to allylic alcohols, notable shortcomings have arisen in the two established methods outlined above. While Simmons-Smith reactivity is regio-, and diastereoselective, it is largely limited to methylenation¹⁰ – substituted methylene transfer remains underdeveloped. Metal-catalyzed diazo decomposition fails with allylic alcohols as conversion to cyclopropanes is low yielding due to competitive O–H insertion.¹¹

We have previously reported that N-enoxyphthalimides are a unique one-carbon component for the cyclopropanation of activated alkenes.¹² Furthermore, tuning the cyclopentadienyl (Cp) ligand on the Rh^III catalyst delivers either cis- or trans-disubstituted cyclopropanes stereoselectively.^{13, 14} In a complementary approach, we found that exchanging trifluoroethanol (TFE) solvent for methanol (MeOH) and again tuning the Cp ligand on the Rh catalyst, activated alkenes undergo syn-1,2-carboamination.¹⁵ This chemodivergence is hypothesized to originate from MeOH participating as a nucleophile to open the phthalimide ring that allows the N-enoxyphthalimide to act as a bidentate ligand throughout catalysis. On the basis of these findings, we sought to expand the scope of our reported cyclopropanation toward unactivated alkenes. Herein, we report that allylic alcohols undergo directed diastereoselective cyclopropanation using this approach.

Initial investigations began with phenyl-N-enoxyphthalimide 1a and trans-2-hexen-1-ol 2a in the presence of various Rh(III) catalysts in TFE at room temperature delivering cyclopropane 3aa in moderate yield but high diastereoselectivities (Table 1, entries 1–4). A solvent (entries 5 and 6) and base screen revealed that KOPiv in TFE is optimal, providing 64% yield and >20:1 d.r. for the desired product (entry 7). Furthermore, we discovered that reducing the reaction temperature to 0 °C leads to the desired cyclopropane in 81% yield while preserving excellent diastereoselectivity (entry 8).

Table 1.

Reaction Optimization^a

Entry	CpX	Base	Solvent	Yield 3aac
1	Cp*	KOAc	TFE	24%
2	Cp*CF3	KOAc	TFE	50%
3	Cp*t-Bu	KOAc	TFE	22%
4	CpE	KOAc	TFE	45%
5	Cp*CF3	KOAc	MeOH	36%
6	Cp*CF3	KOAc	THF	29%
7	Cp*CF3	KOPiv	TFE	64%
8d	Cp*CF3	KOPiv	TFE	81%e

^aConditions: 1a (1 equiv.), 2a (1.2 equiv.), Base (2 equiv.), [Cp^XRhCl₂]₂ (5 mol%), in solvent (0.2M) at 21 °C for 16 hours.

^bDetermined by the ¹H-NMR of the unpurified reaction mixture.

^cYields determined by ¹H-NMR.

^dReactions carried out at 0 °C instead of 21 °C.

^eIsolated yield.

We next examined if the diastereoselectivity of the tri-substituted cyclopropane product was directly correlated with initial alkene geometry (Scheme 2). Both trans- and cis-1,2-disubstituted primary allylic alcohols provide the desired cyclopropanes 3aa and 3ab in good yield–81% and 62%, respectively–and >20:1 d.r., implicating a stereospecific transformation with respect to the alkene. Similar to the parent allylic alcohol, we found crotyl alcohol gives cyclopropane 3ac in excellent diastereoselectivity and 81% yield. Methallyl alcohol gives cyclopropane 3ad in 62% yield with 7.0:1 d.r. while prenyl alcohol furnishes 3ae in 82% yield and >20:1 d.r. To showcase the regio-preference of our cyclopropanation protocol, 1a was subjected to substrate 2f (geraniol) bearing a tethered tri-substituted alkene as a potential competitive site for cyclopropanation. Gratifyingly, cyclopropane 3af was generated in 55% yield with good diastereoselectivity and excellent regioselectivity.

Scheme 2.

Selectivity of Cyclopropanation

With optimized conditions in hand, we examined the scope of this reaction (Scheme 3). Varying para- (3ba-3ea) and meta- (3fa-3ha) arene substitution on the enoxyphthalimide is tolerated, with each substrate displaying >20:1 diastereoselectivity. ortho-Fluorine containing enoxyphthalimide delivers cyclopropane 3ia in 44% yield. An alkyl substituted N-enoxyphthalimide¹⁶ is also a competent substrate, affording cyclopropane 3ka in 92% yield.

Scheme 3.

N-Enoxyphthalimide Scope

Next, a range of suitable allylic alcohols for the cyclopropanation reaction was explored (Scheme 4). Notably, chiral allylic alcohol substrates provide additional complexity leading to the potential of four different stereoisomers. In the event, these reactions deliver the corresponding cyclopropanes 3ag-3ai with varying levels of diastereoselectivity depending on the substituent size, from vinyl (73%, 2.5:1 d.r., major to Σ minor), to methyl (69%, 7.1:1 d.r.) and phenyl (62%, >20:1 d.r.). Using trans-1,2-disubstituted secondary allylic alcohols, we observed single diastereomers of cyclopropanes 3aj-3al in outstanding yields. Finally, we sought to assess the directing ability of cyclic allylic alcohols. When cyclohexenol 2m is subjected to the reaction conditions, no cyclopropane is observed. However, with cyclooctenol 2n, cyclopropane 3an is afforded in excellent yield and diastereoselectivity. Based on previous Simmons-Smith cyclopropanations that explore directing effects of cyclic systems,¹⁷ we hypothesize that the larger ring accomodates the correct C–O bond angle for the cyclopropanation to occur, affording the same relative stereochemistry as observed with acyclic allylic alcohols.

Scheme 4.

Allylic Alcohol Scope

To interrogate the mechanism of this cyclopropanation reaction (Scheme 5), we next subjected 1a to the reaction conditions in the absence of alkene with TFE-d₁ solvent and observed no deuteration of the alkenyl protons suggesting that the C–H activation is irreversible. Homoallylic alcohol 4a gives cyclopropane 5aa in only 12% yield indicating the chain length from the oxygen atom to the olefin is of great importance. Similarily, bishomoallylic alcohol 6a gives cyclopropane 7aa in only 17% yield. Allylic ether 8a is a poor substrate with only trace 9aa observed indicating the presence of an unhindered hydroxyl-group is necessary for the reaction to take place. Allylic carboxylic acid 8b gives cyclopropane 9ab in trace yield. Interestingly, protected allylic ether 8c gives cyclopropane 9ac in 77% yield and 9.5:1 d.r. indicating the importance of the strength of the pendent nucleophile. In another experiment, we set out to detect potential reactivity between 1a and 2b in the absence of Rh catalyst and we were surprised to observe the formation of dioxazoline 10ac in 38% yield with 1 equivalent of KOPiv in THF at room temperature. We speculate this occurs via opening of the phthalimide ring and acylation of the allylic alcohol (eq. 8). Subjecting dioxazoline 10ac to the cyclopropanation reaction conditions did not afford cyclopropane, suggesting that dioxazoline 10ac is an off-cycle product.

Scheme 5.

Mechanistic Experiments

On the basis of these experiments, we propose the following mechanism (Scheme 6). First, 2 undergoes acylation with 1 that gives intermediate I. Maintaining the reaction temperature at 0 °C inhibits cyclization of 8ab, which is instead intercepted by the active Rh(III) catalyst II. Intermediate I undergoes irreversible C–H activation via concerted metalation-deprotonation that gives rhodacycle III. At this stage, we hypothesize the formation of intermediate IV by cleavage of the N–O bond and formation of a Rh-carbenoid. Due to the prior acylation of the allylic alcohol, intermediate V is formed via the [2+1] annulation where the Rh-carbenoid is delivered across the alkene and on the same face as the pendent oxygen atom in stereoselective fashion. Protodemetallation and subsequent phthalimide ring closure releases the product and turns the catalyst over.

Scheme 6.

Proposed Catalytic Cycle

In summary, we have developed a directed diastereoselective cyclopropanation protocol for the [2+1] annulation of N-enoxyphthalimides and allylic alcohols. The diastereoselectivity of the reaction is speculated to arise from an intermediate generated by a ring-opening acylation of the allylic alcohol. Generation of a Rh-carbenoid leads to intramolecular cyclopropanation in excellent yield and diastereoselectivity.

Scheme 1.

Directed Cyclopropanations and Precedent Involving N-Enoxyphthalimides