Gold(I)-Catalyzed Enantioselective Ring Expansions of Allenyl Cyclopropanols and Cyclobutanols

Abstract

Herein we report enantioselective gold(I)-catalyzed ring expansions of allenyl cyclopropanols and allenyl cyclobutanols. Cyclobutanones and cyclopentanones bearing a vinyl-substituted quaternary chiral centers are formed in good yields and with excellent enantiomeric excesses. Efficient chiral inductions are achieved via asymmetric gold-ligand cooperation, which is enabled by chiral bifunctional phosphine ligands. This reaction employs a low catalyst loading and proceeds under mild reaction conditions. With a substrate possessing an additional prochiral center, an additional chiral center was set up with excellent diastereoselectivity, demonstrating the synthetic utility of this chemistry in accessing structures of increasing stereochemical complexity.

Graphical Abstract

For the past several years, we have developed various bifunctional biaryl-2-ylphosphine ligands¹ possessing a remote Lewis/BrØnsted basic group for the implementation of metal-ligand cooperation² in gold catalysis.³ By leveraging the linear bis-coordinated geometry of gold(I) complexes – a feature proving to pose challenges to asymmetric gold catalysis,⁴ these ligands form H-bond interactions with substrates/incoming nucleophiles via their remote basic group. By employing their chiral variants, our group has documented a new strategy for achieving enantioselective gold(I) catalysis via asymmetric gold-ligand cooperation and developed several enantioselective methodologies including dearomatization,⁵ asymmetric cyclization/cycloisomerization,⁶ asymmetric protonations,⁷ asymmetric trapping of in-situ generated reactive intermediates,⁸ and desymmetric ring expansion of alkynylcyclobutanols.⁹ In the last study, as shown in Scheme 1A, the selective migration of one of the ring C-C bonds in the stereogenic ring-expansion step is enabled by a key hydrogen bonding interaction between a ligand basic group and the substrate hydroxyl group, as depicted in structure A. This interaction controls the rotation of the C(sp)-C(sp3) bond shown in dark red and preferentially orients one of the ring C(sp3)-C(sp3) (shown in dark green) antiperiplanar to the gold-alkyne interaction, allowing for its selective migration.¹⁰

Scheme 1:

Asymmetric ring expansion reactions: precedents and design

As a continuing exploration of our gold-ligand cooperation strategy in asymmetric ring expansion, we set our sights on allenylcycloalkanol substrates and in particular allenylcyclopropanols and allenylcyclobutanols. The first asymmetric ring expansion of allenylcyclopropanols was reported by Toste in 2009,¹¹ wherein cyclobutanones bearing an all-carbon quaternary center were synthesized with good to excellent enantioselectivities in the presence of a chiral BIPHEP(AuCl)₂ precatalyst (Scheme 1B). In the cases of allenylcyclobutanols, Trost and coworkers employed allenyl ether-based cyclobutanols¹² as substrates and achieved enantioselective and diastereoselective Wagner-Meerwein shifts via asymmetric Pd catalysis, furnishing functionalized cyclopentanones bearing an O-substituted quaternary chiral center (Scheme 1B).¹³ We envisioned that our cooperative gold catalysis strategy might also be conducive to enantioselective ring expansion of allenylcycloalkanols, wherein cationic gold could activate allene to promote ring expansion while the chiral induction could be realized by the hydrogen bonding interaction between the substrate HO group and a ligand remote functional group, as shown in structure B in Scheme 1C.

We first employed the phenyl-substituted allene-cyclopropanol 1a as the substrate for reaction discovery and conditions optimization. NaBARF was used in excess as a chloride scavenger to activate precatalyst LAuCl throughout the process, and the results are shown in Table 1. With the ligand (S)-L1, which features a methyl-substituted chiral center at the C1 position of the tetrahydroisoquinoline moiety, the gold-catalyzed ring expansion reaction delivered α-vinyl-α-phenylcyclobutanone (2a) in excellent yield, albeit the enantioselectivity was low (10% ee, entry 1). The use of analogous ligand (S)-L2, which features a cyclohexyl group at the C1 position of the ligand tetrahydroisoquinoline ring resulted in a similarly high yield, but 2a was isolated as a racemate (entry 2). We then screened several chiral binol-derived ligands (L3–L7)^5a for this transformation. These bifunctional ligands feature an amide (L3–L4), a phosphine oxide (L5–L6), or a phosphonate moiety (L7) at the C3’ position. Pleasingly, 2a was formed with moderate to excellent enantiomeric excesses (entries 3–7). The phosphonate ligand L7 was the most effective in asymmetric induction, and 2a was formed with 90% ee at ambient conditions (entry 7). The product ee value was improved to 93% upon lowering the reaction temperature to 0 °C (entry 8). PhCF3 was screened as a solvent but showed slightly diminished yield and enantioselectivity (entry 9). Lowering the precatalyst loading by 10 fold (i.e., from 5 mol% in entry 8 to 0.5 mol % in entry 10), led to slightly improved product yield and ee value, albeit requiring much longer reaction time (16 h). The (S)-configuration of 2a was established by comparing its chiral HPLC data to those published.¹¹ The absolute stereochemistries of the products in the scope table (Table 2) are assigned analogously.

Table 1:

Reaction discovery and conditions optimization

entry:	ligand:	x:	y:	solvent:	temp.:	time (h):	yield (%):	eeb (%):
1	L1	5	10	DCM	rt	16	90	10
2	L2	5	10	DCM	rt	16	95	0
3	L3	5	10	DCE	rt	0.1	84	86
4	L4	5	10	DCE	rt	0.1	71	84
5	L5	5	10	DCE	rt	0.1	56	67
6	L6	5	10	DCE	rt	0.75	85	89
7	L7	5	10	DCE	rt	0.1	73	90
8	L7	5	10	DCM	0 °C	0.75	92	93
g	L7	5	10	CF3Ph	0 °C	0.75	83	90
10	L7	0.5	1.5	DCM	0 °C	16	95	96

^aReaction was performed at 0.050 mmol scale with yield determined by NMR using diethyl phthalate as the interal standard.

^bDetermined by HPLC equipped with chiral stationary phase-equipped column.

^cReaction was performed at 0.25 mmol scale, isolated yield reported. In call cases 100% conversion from starting material was observed.

Table 2:

Scope of allenyl cyclopropanols

^aReaction performed using 5 mol % L1AuCI and 10 mol % NaBARF.

^bReaction performed using 1 mol % L1AuCI and 2 mol % NaBARF.

With the optimized conditions in Table 1, entry 10 established, we explored the scope of this chemistry. As shown in Table 2, the reaction worked well with aryl-substituted allene-cyclopropanol substrates. Various halide substitutions on the benzene ring of 2a, such as p-F (2b), p-Cl (2c), and m-Br (2d) were readily accommodated, and the reactions exhibited excellent enantioselectivities. Surprisingly, the reaction leading to 2e possessing an o-bromophenyl group exhibited only 40% yield and 85% ee under the optimized conditions. We attributed the poor results to steric hindrance posed by the o-bromo group. The tertiary amine ligand L1 (5 mol %), however, led to a more enantioselective and efficient reaction, delivering the product in 95% yield and 89% ee. Electron-donating substituents on the phenyl ring including 4-Me (2f) and 4-OMe (2g) were also conducive to good yields and excellent enantioselectivities in this ring-expansion reaction. In the case of the substrate bearing a p-CF₃, (S)-L1 is the optimal ligand. Due to a competitive 5-endo-trig cyclization leading to a dihydrofuran side product, the yield of 2h is moderate, although its enantiomeric excess (96% ee) was excellent. Other aryl groups such as 2-naphthyl and 2-thiophenyl were readily accommodated, and the corresponding products 2i and 2j were formed in good yields and with excellent enantioselectivities. Our attempts to expand the reaction scope to alkyl-substituted allene substrates were unsuccessful, with the reactions exhibiting diminished yields (63% to 79% yield) and moderate enantioselectivities (40% to 82% ee) (see Supporting Information for more details).

We then shifted our efforts to the ring expansion of allenyl cyclobutanols. Initially, substrates featuring aryl/alkyl-substituted allenes were tested. Disappointingly, 5-endo-trig cyclizations forming dihydrofurans proved to be exclusive, and none of the desired ring expansion was observed. This phenomenon is consistent with Shin’s report.¹⁴

Inspired by Trost’s ring expansion of allenyl ether-cyclobutanols,¹³ we surmised that a more electron-rich allenyl ether moiety might favor ring expansion over competing 5-endo-trig cyclization upon gold(I) activation. This consideration was corroborated by a single racemic case reported by Shin.¹⁴ However, to our disappointment, when we attempted to synthesize 4a from the corresponding allenyl ether substrate 3a, the reaction using L7 as the ligand again yielded predominantly a dihydrofuran product. Much to our delight, with the chiral tertiary amine ligand (S)-L1, we achieved chemoselectivity favoring the ring expansion under ambient conditions, affording the chiral cyclopentanone product in 86% yield and with an outstanding 92% ee (Table 3). We attribute this contrast in chemoselectivity to the distinctive differences in the hydrogen bonding interaction between the substrate and the catalyst remote functional group. Of note is that the catalyst loading can be as low as 1 mol %. We subsequently studied the scope of this ring expansion by varying the allene alkoxy/aryloxy substituent. A phenoxy group permitted a smooth reaction, affording the cyclopentanone 4b in a moderate yield and with 95% ee. Various benzyloxy groups including PMB (4c), 4-bromobenzyl (4d), and 2,6-dichlorobenzyl (4e) were allowed. In the case of PMB, the ee value (85%) was notably lower than in the other two cases, and a higher catalyst loading (5 mol %) was needed. In addition, a naphthyl-1-methyl (4f) and a thiophenyl-2-methyl (4g) were readily accommodated as the ethereal substituent, affording the corresponding chiral cyclopentanones in excellent yields and with exceptional enantiomeric excesses. The absolute configuration of 4c is assigned by comparing its optical rotation to that reported by Trost.^13b The stereochemistry of the other cyclopentanone products were assigned analogously. We also examined the scalability of this transformation by running a 1 mmol-scale reaction of 3a. and 4a was isolated in a comparable yield (87%) but with a lower 87% ee.

Table 3:

Scope of allenyl cyclobutanols

^aWith 1 mmol of 3a, both the yield and ee of 4a were 87%.

^bReaction was performed using 5 mol % L1AuCI and 10 mol % NaBARF

Our attempts to expand the scope of this ring expansion reaction to the analogous allenyl ether-cyclopentanol substrates were again thwarted by facile 5-endo-trig cyclizations, leading to undesired dihydrofuran products.

We also examined the reaction of the allenyl ether-cyclobutanol substrate cis-5 containing a prochiral center on the cyclobutane ring (Eq 1). Its preparation and the structural assignment were based on the previous report by Trost.^13b With (S)-L1 as the ligand, cyclopentanone 6 was formed in 92% yield and with 97% ee. The absolute configurations of the product were established by comparing its optical rotation to that reported in the literature.^13b Moreover, the diastereoselectivity (~20:1) is excellent. In comparison, the asymmetric palladium catalysis developed by Trost^13b exhibited 88% yield, 88% ee for the major diastereomer, and a diastereomeric ratio of ~10:1.

Eq 1

Scheme 2 outlines two stereochemical models to rationalize the reaction enantioselectivities. In the reaction of allenylcyclopropanols, the H-bond between the phosphonate moiety of (S)-L7 and the substrate HO group helps position the Si face of the allene internal C-C double bond for gold coordination, as shown in the structure C. The following migration/ring expansion of the antiperiplanar ring C-C bond (colored in dark blue) leads to the observed (S)-configuration of the cyclobutanone products 2. During the reaction of allenylcyclobutanols, a similar two-point binding between L1Au⁺ and the substrate is proposed in D, the transformation of which to the cyclopentanone products 4a-4g establishes the (R)-configuration at the ring quaternary center. In the case of 5, this two-point binding also orients the dark green-colored bond of the cyclobutane ring antiperiplanar to gold-allene interaction and hence permits its selective migration. As such, 6 possessing the shown tertiary chiral center is formed with excellent diastereoselectivity.

Scheme 2:

Proposed models rationalizing the reaction stereochemical outcomes

In conclusion, we developed an efficient asymmetric ring expansion of allenylcyclopropanols and allenyl ether-cyclobutanols, furnishing cyclobutanones and cyclopentanones featuring a vinyl-substituted quaternary center with mostly excellent enantiomeric excesses. Excellent diastereoselectivity was realized with a substrate featuring an additional prochiral center, revealing synthetic utilities of this approach in accessing structures of increasing stereochemical complexity. Low catalyst loadings (0.5 – 1 mol %) are allowed, and the reaction conditions are exceedingly mild. This work represents a further demonstration of the potential of the ligand-enabled gold-ligand cooperation strategy in achieving synthetically versatile asymmetric transformation.

Supplementary Material

Experimental procedures, compound characterization data, NMR spectra, and chiral HPLC chromatographs. This material is available free of charge via the Internet at http://pubs.acs.org.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.