Desymmetrization of meso-Dibromocycloalkenes through Copper(I)-Catalyzed Asymmetric Allylic Substitution with Organolithium Reagents.

The highly regio- and enantioselective (up to >99:1 dr, up to 99:1 er) desymmetrization of meso-1,4-dibromocycloalk-2-enes using asymmetric allylic substitution with organolithium reagents to afford enantioenriched bromocycloalkenes (ring size of 5 to 7) has been achieved. The cycloheptene products undergo an unusual ring contraction. The synthetic versatility of this Cu(I)-catalyzed reaction is demonstrated by the concise stereocontrolled preparation of cyclic amino alcohols, which are privileged chiral structures in natural products and pharmaceuticals and widely used in synthesis and catalysis.

The enantioselective desymmetrization of meso compounds is one of the most powerful strategies in organic synthesis.[1] It enables the formation of compounds with multiple stereocenters in a single step from readily accessible σ-symmetric precursors. In the case of meso-cycloalk-2-ene-1,4-diol derivatives, desymmetrization by asymmetric allylic substitution (AAS) is a powerful tool for the construction of enantiomerically enriched functionalized cyclic products,[2] which have found ample use in the total syntheses of various natural products.[3] Depending on the choice of nucleophile (soft or hard) and metal catalyst, the reaction can result in either α- or γ-substitution, with either retention or inversion of configuration. The most commonly employed procedure is the Pd-catalyzed desymmetrization, which is usually performed with soft nucleophiles to give SN2 products (Scheme a).[2,3] A viable alternative is the Rh-catalyzed desymmetrization using arylboronic acids,[4] which give SN2 or SN2′ products depending on the ligand at Rh. These processes, albeit highly versatile at producing chiral building blocks, rely on precious metal catalysts. In contrast, there are markedly few examples of the Cu(I)-catalyzed desymmetrization, which generally employs hard nucleophiles to provide SN2′ products.[5] Sawamura and co-workers have utilized the Cu-catalyzed asymmetric boryl substitution in conjunction with allylation to afford a formal SN2 substitution with electrophiles.[6]

Scheme 1

Desymmetrization of meso-1,4-Cycloalkenediol Derivatives

The Cu(I)-catalyzed AAS with organometallic nucleophiles, pioneered by Bäckvall and van Koten in 1995,[7] is an effective method to synthesize tertiary carbon stereocenters.[8] While many different metal catalysts and organometallic nucleophiles could be used for AAS,[9] the readily available organolithium reagents were considered too reactive to be utilized in catalytic asymmetric C–C bond formation until the 2011 disclosure by Feringa et al. using allylic bromides as substrates, forming SN2′ products with high regio- and enantioselectivities.[10] In recent years our group has extended this protocol,[11] most notably to the use of allylic-chlorides and -ethers[11a,11b] and aryllithium nucleophiles[11c,11d] and also to the formation of highly challenging all-carbon quaternary stereocenters.[11b,11d] We envisaged that the AAS strategy with organolithium reagents could be applied to the desymmetrization of meso compounds. Herein, we report the highly regio- and enantioselective (up to >99:1 dr, up to 99:1 er) desymmetrization of meso-2-cycloalkene-1,4-dibromides using Cu(I)-catalyzed AAS with organolithium reagents to afford enantioenriched bromocycloalkene synthons (Scheme b).

Optimization of the desymmetrization reaction began with meso-3,6-dibromocyclohex-1-ene 1 as model electrophile and commercially available n-BuLi as nucleophile in the presence of a catalytic amount of CuBr·SMe2 and chiral ligand. The racemic reaction with PPh3 as ligand (Table , entry 1) proceeded to full conversion to give trans-4-bromo-3-butylcyclohexene 2d as the major product (from SN2′ substitution) in 91% yield. The double addition product 3 (9%) was also observed; its formation most probably occurs via a SN2-type substitution followed by a SN2′-type substitution on the allylic bromide intermediate. Taniaphos L1, which was an effective chiral ligand in the acyclic AAS,[10] was initially tested (entry 2). Unfortunately, no conversion was observed that (based on models) was attributed to steric interactions between L1 and cyclohexene 1. We then switched to the phosphoramidite ligand class,[12] which has previously been used in the desymmetrization of meso-cyclic bis(diethyl phosphates) by Cu-AAS using organozinc reagents.[5b,5c] With (S,R,R)-phosphoramidite L2, only partial conversion was observed, and the desired product had low er (entry 3). When (S,S,S)-phosphoramidite L3 was tested, 90% conversion (98:2 er) to the desired product was found (entry 4). When this transformation was performed on multigram scale, analytically pure 2d was obtained in 89% yield and 99:1 er. Neither a more electron-rich phosphoramidite L4 nor a more flexible octahydrophosphoramidite L5 could enhance this result (entries 5 and 6). When a 9:1 cis/trans mixture of starting material was subjected to the optimized conditions with L3, the enantioselectivity was maintained (99:1 er), and the product 2d could be isolated in 80% yield (entry 7); trans-1 was almost entirely recovered. This prompted us to investigate the reaction with racemic trans-1 under the same conditions (entry 8). Unsurprisingly, the reaction did not proceed to full conversion, and formation of some cis-4-bromo-3-butylcyclohex-1-ene 4 was also observed. The absolute configuration of 2d was determined by X-ray crystallography of diol 5 (Table , inset),[13] resulting in a Flack parameter of x = 0.04(2). Chiral HPLC confirmed that a single diastereomer of 5 with four contiguous stereocenters was obtained (>99:1 dr, 99:1 er) after Upjohn dihydroxylation of 2d.

Table 1

Screening of Ligands for AAS-Desymmetrization of meso-Dibromocyclohexene 1 with n-BuLia

Conditions: meso-1 (0.2 mmol) in CH2Cl2 (2 mL). n-BuLi (0.24 mmol, 1.6 M solution in hexanes diluted to a final concentration of 0.24 M) was added over 2 h.

Determined by GC–MS and 1H NMR.

er determined by chiral GC.

A 9:1 cis/trans mixture of 1 was used.

Isolated yield of 2d on 0.2 mmol scale; increases to 89% on 10 mmol scale (see SI).

Racemic trans-1 was used. Inset: Ball-and-stick representation of the X-ray crystal structure of diol 5.

With the optimized conditions in hand (entry 4), we proceeded to examine the scope of the reaction. Continuing with the six-membered substrate 1 (Scheme ), the addition of commercially available alkyllithium reagents afforded the AAS products 2a–e with excellent enantioselectivities (up to 99:1 er). Only isopropyl-bearing product 2c had a slightly lower er (95:5), possibly a result of the steric bulk of the isopropyl group. The reaction worked similarly well for meso-3,5-dibromocyclopentene 6 to generate products 7a–e in good yields with up to 96:4 er (Scheme ).

Scheme 2

Alkyllithium Scope for Desymmetrization of Five- and Six-Membered meso-Cyclic Allylic Dibromides 1 and 6,,

Conditions: meso-1 (9:1 cis/trans) or 6 (0.2 mmol) in CH2Cl2 (2 mL). RLi (0.24 mmol, diluted to a final concentration of 0.24 M) was added over 2 h.

Isolated yields.

er determined by chiral GC.

GC yields reported due to product volatility (see SI).

When meso-3,7-dibromo-cycloheptene 8 was used in the desymmetrization reaction with alkyllithium reagents (Scheme ), the expected products 9a–e (>99:1 dr) were initially obtained with er values ranging from 90:10 to 97:3, based on NMR and chiral GC. However, when purification of these seven-membered rings 9a–e was attempted by flash column chromatography on silica, only their corresponding cyclohexene analogs 10a–e were isolated with complete stereospecificity. A detailed structural analysis and mechanistic and theoretical study to elucidate this remarkable ring contraction are reported separately.[14]

Scheme 3

Desymmetrization-Rearrangement of Seven-Membered meso-Cyclic Allylic Dibromide 8,,

Conditions: (i) meso-8 in CH2Cl2 (2 mL). RLi (0.24 mmol, diluted to a final concentration of 0.24 M) was added over 2 h; (ii) silica, pentane.

Isolated yields.

er of 9a–e and 10a–e determined by chiral GC to be the same, so enantiospecificity > 99%.

We hypothesized that a phenyl substituent would stabilize the desymmetrization product, i.e., chiral cycloheptene 9, enabling its isolation. We have previously reported that N-heterocyclic carbenes (NHC) are the most suitable ligand class for asymmetric allylic arylation (AAAr).[11c,11d] As such, we screened, besides achiral L6 as control, several chiral NHC ligands for the desymmetrization of dibromocycloheptene 8 with phenyllithium (Table ). While the dihydroimidazolium-based ligands L7 and L8 gave excellent conversion, the er was poor to moderate (entries 2 and 3). In contrast, triazolium-based ligands L9 and L10 gave poorer conversions (entries 4 and 5). Gratifyingly, we found that imidazolium salt L12 was a suitable NHC precursor; in conjunction with CuBr·SMe2 and NaOt-Bu, this catalytic system afforded the desired 4-bromo-3-phenylcycloheptene 9f in 83% isolated yield with 95:5 er (entry 7). In accordance with our prediction, and in sharp contrast with alkyl analogs 9a–e, this product was stable to base-treated silica and could be isolated. The absolute configuration of 9f was determined to be (R,R) by X-ray crystallography of diol 11 (Table , inset),[15] which was obtained via diastereoselective Upjohn dihydroxylation (88:12 dr, 96:4 er as determined by chiral HPLC).

Table 2

Screening of Ligands for AAAr-Desymmetrization of meso-Dibromocycloheptene 8 with PhLia

Conditions: meso-8 (0.2 mmol) in CH2Cl2 (2 mL). PhLi (0.30 mmol, 1.9 M solution in di-n-butyl ether diluted with hexanes to a final concentration of 0.30 M) was added over 2 h.

Determined by GC–MS and 1H NMR.

er determined by chiral GC.

Isolated yield of 9f. Inset: Ball-and-stick representation of the X-ray structure of diol 11.

Cyclic amino alcohols are structural elements found in numerous natural products, e.g., tropane alkaloids,[16] and are privileged scaffolds in medicinal chemistry, e.g., atropine and cocaine.[18] Having access to a variety of enantioenriched bromocycloalkenes of various ring sizes via the AAS-desymmetrization protocol, we next demonstrated the versatility of these products by the concise stereocontrolled synthesis of cyclic amino alcohols (Scheme ). Reaction of cyclohexene 2d with m-CPBA afforded a 71:29 diastereomeric mixture of epoxides. Ring opening of the epoxide with benzylamine catalyzed by silica under neat conditions was selective for the major epoxide isomer, affording trans-1,2-aminoalcohol derivative 12 in 60% yield over two steps. SN2 substitution of bromide 12 with sodium azide followed by hydrogenation yielded trans-1,4-diamino-2-alcohol 13 with four contiguous stereocenters (Scheme a). The seven-membered analog cycloheptene 9f undergoes diastereoselective Upjohn dihydroxylation (88:12 dr) to afford cis-1,2-diol 11 in 80% yield, which was readily transformed into aminodiol 14 via substitution and hydrogenation (Scheme b).

Scheme 4

Derivatization of Desymmetrization Products Towards Cyclic Aminoalcohols

Conditions: (i) m-CPBA (1.2 equiv), PhMe, RT; (ii) BnNH2 (1.2 equiv), silica (10 wt %), 80 °C; (iii) NaN3 (3 equiv), DMF, 80 °C; (iv) H2 (1 atm), Pd/C (20 mol %), EtOAc; (v) OsO4 (4 mol %), NMO (1.5 equiv), acetone/H2O (3:1).

Aminodiol 14 is a direct precursor to 2-phenyl-tropan-6α-ol using the cyclization strategy described by Pollini et al.[17] These 8-azabicyclo[3.2.1]octanes[18] represent an important scaffold of bioactive tropane alkaloid natural products such as schizanthines, baogongtengs, and calystegines.[16b,19] Thus, our synthesis of aminodiol 14 represents an efficient route to phenyl-substituted analogs of these natural products and drug targets (see Figure ).

Figure 1

Examples of tropane alkaloids with the 8-azabicyclo[3.2.1]octane framework.

In summary, the highly regio- and enantioselective desymmetrization of meso-dibromocycloalkenes with ring size ranging from 5 to 7 via Cu-AAS with organolithium reagents has been demonstrated. Phosphoramidite L3 is the preferred ligand for alkyllithium reagents, while for arylation NHC was found to be the ligand of choice. These findings represent an efficient method to access enantioenriched cyclic bromoalkenes; the synthetic utility of the products is demonstrated by the concise synthesis of chiral multifunctional cyclic aminoalcohols, which are a privileged scaffold for natural products, pharmaceuticals, and asymmetric synthesis.