Chiral Bifunctional Phosphine Ligand-Enabled Cooperative Cu Catalysis: Formation of Chiral α,β-Butenolides via Highly Enantioselective γ-Protonation

Abstract

α,β-Butenolides with ≥96% enantiomeric excess are synthesized from β,γ-butenolides via a novel Cu(I)-ligand cooperative catalysis. The reaction is enabled by a chiral biphenyl-2-ylphosphine ligand featuring a remote tertiary amino group. DFT studies support the cooperation between the metal center and the ligand basic amino group during the initial soft deprotonation and the key asymmetric γ-protonation. Remarkably, other coinage metals, i.e., Ag and Au, can readily assume the same role as Cu in this asymmetric isomerization chemistry.

Graphical Abstract

Many natural products featuring chiral α, β-butenolide motifs possess various biological activities (Scheme 1A). For example, avenolide can control the production of antibiotics in Streptomyces avermitilis,¹ thorectandrols B inhibits the growth of MALME-3M (melanoma) and MCF-7 (breast) cancer cell lines,² and kallolide A exhibits anti-inflammatory activity.³ Various synthetic approaches have been developed to access chiral α,β-butenolide.⁴ The direct catalytic olefin isomerization of racemic/achiral β,γ-butenolide into chiral α,β-butenolide via the achiral 2-furanoxyl anion A is an atom-economic and arguably the simplest approach (Scheme 1B), yet has been only sparsely explored. A notable advance in this strategy was achieved by Deng⁵ in 2011 by employing a cinchona-derived organocatalyst. In this chemistry, good levels of enantioselectivities (87–94% ee) are achieved for γ-monosubstituted and α, γ-disubstituted α, β-butenolide products, but only moderate ee values (81–82%) are reported for the β,γ-disubstituted α,β-butenolides. This chemistry was applied in the total synthesis of Leucosceptroid family of natural products,⁶ where a moderate diastereomeric ratio of 7/1 was reported with the latter butenolide type. Herein, we report a Cu(I)-bifunctional phosphine cooperative catalysis that achieves the asymmetric butenolide isomerization which exhibits enantiomeric excess ranging from 96% to 99%. Moreover, this chemistry is applicable to all of the three aforementioned substitution patterns and proceeds at ambient temperature and mostly in 0.5 h.

Scheme 1.

Butenolide natural product and reaction design

For the past several years, we⁷ have developed various bifunctional biphenyl-2-ylphosphine ligands⁸ featuring a remote basic group for gold-ligand cooperative catalysis.^8–9 Scheme 1C shows the chiral ligands that have enabled asymmetric cooperative gold catalysis.^8a,8c,8e Among them, L2 and L3 feature a fluxional biphenyl axis and a remote tertiary amino group possessing an α-chiral center and engaging in critical propargylic deprotonation during catalysis, as revealed by the DFT-optimized TS B using (S)-L3 as the ligand.^8e Despite the syn-periplanar nature of the concerted Au-activation and amino-deprotonation, we envisioned that an orthogonal organization of a ‘pulling’ cationic metal and a ‘pushing’ basic amino group can be readily achieved by these chiral ligands, as outlined in the structures C and C’ in Scheme 1D, which should be suited for soft enolization and the reverse protonation. This protonation of the metal enolate could be extended to the γ-protonation of the metal 2-furanolate in D. As such, we envision that chiral ligands such as L2 or L3 might enable highly enantioselective isomerization of β,γ-butenolides via sequential soft enolization and asymmetric γ-protonation.¹⁰ Since Cu^I and Ag^I can also adopt the same linear bis-coordinated structures with bulky phosphine,¹¹ we anticipate that the corresponding Cu^I or Ag^I complexes featuring these bifunctional ligands could also be effective in this cooperative catalysis manifold.¹² Moreover, these harder and cheaper coinage metals may be advantageous in enolate chemistry over softer Au since the coordination/activation of hard carbonyl oxygen instead of soft C-C triple bond is desired. It is noteworthy that metal-ligand cooperative catalysis involving Cu^I13 or Ag^I is scarce,, and this design, if implemented, would constitute the first application of this class of bifunctional ligands in Cu or Ag catalysis.

Guided by these considerations, we initiated our investigation by employing Cu as the metal and the β, γ-butenolide 1a – prepared from an allenic ester in two steps⁵ – as the model substrate. As expected, the Cu^I salt, [Cu(MeCN)₄]⁺ PF₆⁻, alone could not promote the isomerization of 1a into the α,β-butenolide product 2a to a noticeable extent (entry 1), nor was its combination with Et₃N (entry 2). [JohnPhosCu(MeCN)]⁺ PF₆^-11b and Et₃N (5 mol % each) did lead to substantial conversion in 16 h at ambient temperature, albeit in 6% yield (entry 3). The reaction was, however, drastically improved when JohnPhos was replaced by the achiral tertiary amine-functionalized ligand L4.^8b With the in-situ generated L4Cu⁺, the reaction proceeded to completion in 3 h and afforded 2a in nearly quantitative yield (entry 4). This large enhancement of reactivity is consistent with the intended Cu-ligand cooperation. When the chiral ligand (S)-L2 along with [Cu(MeCN)₄]⁺ PF₆⁻ was employed, the reaction again proceeded with excellent efficiency and moreover, the ee of 2a was 98% (entry 5). The (R)-configuration of 2a is inferred by comparing the specific optical rotations of its homologs 2b and 2c (see Table 2) with the literature data.⁵ The replacement of (S)-L2 with (S)-L3^8e led to 94% ee (entry 6). To establish the structure of the Cu(I) catalyst, we prepared [(S)-L2Cu(MeCN)]⁺ PF₆⁻ by following the related protocol for [JohnPhosCu(MeCN)]⁺ PF₆⁻.^11b With this preformed chiral cationic Cu(I) complex as the catalyst, the reaction time was shortened to 1 h, while the yield and ee remained excellent (entry 7). Performing the reaction under argon atmosphere further shortened the reaction time to 30 min (entry 8). This observation is consistent with Cu(I) catalysis as atmospheric oxygen might oxidize Cu(I) to likely nonreactive Cu(II). To further characterize the Cu(I) catalyst, we obtained its single crystals for X-ray diffraction studies. However, the solved structure, as shown in Figure 1, is a dimeric (S)-L2Cu(I) complex with the two-metal center bridged by a molecule of water. Nevertheless, it confirms that the Cu(I) center is bis-coordinated, with the angles of P-Cu-O being 167.8° and 163.2°, respectively. This structural feature supports our reaction design. Moreover, this dimeric complex is equally effective as the catalyst (entry 9). In the control experiments, the ligand itself was not competent (entry 10), and Cu(II) salts such as Cu(OTf)₂ and Cu(hfac)₂ could not serve as the copper source (entry 11). Interestingly, this catalytic system worked equally well with the other coinage metals. Hence, with Ag^I or Au^I at the metal center, nearly identical results were obtained (entries 12 and 13). This interchangeability among the coinage metals is remarkable and rare. The scalability of this Cu(I) catalysis was demonstrated on a gram-scale reaction in entry 14. With 1 mol % of the catalyst, the reaction was completed in one hour and delivered 0.99 g of 2a in 92% yield and with 99% ee.

Table 2.

Reaction Scope ^a

^aReaction was performed in 2-dram sealed vials under argon at room temperature. Yield calculated based on the conversation was included in parentheses. Reaction scale is 0.3 mmol and reaction time is 0.5 h if not specified.

^bThe result reported in ref. 5.

^cReaction was performed at 0.15 mmol scale.

^d5 mol% (R)-L2Cu(MeCN)PF₆ was used.

^eReaction was performed at 40 °C.

Figure 1.

CYL drawing of the dimeric Cu(I) complex. The crystal solvent molecules, i.e., one MeCN and three Et₂O, are omitted for clarity. ∠P1-Cu1-O = 163.2° and ∠P2-Cu2-O = 167.8°

With the optimized reaction conditions (i.e., Table 1, entry 8) in hand, we set out to explore the reaction scope. As shown in Table 2, a series of α, γ-disubstituted α,β-butenolides (2b-2o) were synthesized in yields ranging from 88% to 99% and with ≥96% ee. The R¹ group in this series can accommodate methyl (2b), isopropyl (2c), bulky t-butyl group (2d), and various functional groups including C-C double bonds (2e and 2f), phenyl (2g), thiophen-2-yl (2h), chloro (2i), phenyloxy (2j), and phenylthio (2k). From the substrate prepared from (S)-β-citronellol, (5S, 2’S)-2l was formed with 99% diastereomeric excess when (S)-L2 was employed. By switching the chiral ligand to its enantiomer, the diastereomer (5R, 2’S)-2l was formed with the same level of excellent diastereoselectivity. This ligand-enabled diastereomeric divergence permits flexible and selective access to stereochemical arrays. Little impact on the reactivities was noticed when the R³ group was switched from methyl (2a) to allyl (2m), prenyl (2n), or benzyl groups (2o).

Table 1.

Reaction Condition Optimization

Entrya	Catalyst	Time (h)	Conv.	Yield (%)b	ee (%)c
1	[Cu(MeCN)4]+ PFe−	3	2	NA	NA
2	[Cu(MeCN)4]+ PF6− /Et3N	3	3	NA	NA
3	[JohnPhosCu(MeCN)]+ PF6−/Et3N	16	35	6	NA
4	[Cu(MeCN)4]+ PF6− /L4	3	100	99	NA
5	[Cu(MeCN)4]+ PF6− /(S)-L2	3	100	99	98
6	[Cu(MeCN)4]+ PF6− /(S)-L3	3	100	99	94
7	[(S)-L2Cu(MeCN)]+ PF6 −	1	100	99	97
8 d	[(S)-L2Cu(MeCN)]+ PF6−	0.5	100	99 e	98
9d,f	{[(S)-L2Cu]2(H2O)}2+ (PF6−)2	0.5	100	99	98
10	(S)-L2	3	NA	NA	NA
11	Cu(OTf)2 or Cu(hfac)2/(S)-L2	24	<5	<5	NA
12	(S)-L2AuCl/ NaBArF4 (10 mol%)	0.5	100	99	99
13	[Ag(MeCN)2]+ BArF4− /(S)-L2	0.5	100	99	98
14g	[(S)-L2Cu(MeCN)]+ PF6−	1	100	92h	99

^aReaction was performed at 0.05 mmol scale in 1-dram vials.

^bThe NMR yield is calculated by assuming that the triplet at around 0.87 ppm corresponds to the terminal methyl groups of all compounds.

^cDetected using chiral HPLC.

^dUnder Ar atmosphere.

^eIsolated yield.

^fReaction was performed with 2.5 mol% catalyst.

^gReaction was performed with 1 mol % of catalyst.

^h0.99 g product isolated.

Next, we turned our attention to the synthesis of β, γ-disubstituted α, β-butenolides (2p-2s). Much to our delight, they were also formed with ≥97% ee. The β-substituent, i.e., R², can be sterically demanding isopropyl (2q) or part of 6-/7-membered ring connected to the γ substituent (2r and 2s).

Finally, the preparation of chiral monosubstituted α, β-butenolides (2t and 2u) was examined. As expected, in the absence of α- and/or β-substituents to stabilize the product double bond, the energy differences between the substrates and the products appear to be small. As such, this asymmetric isomerization was sluggish and could not reach full conversion due to reaction equilibrium. 40 °C and 17 h were employed to improve the reaction yields without compromising the exceptional enantioselectivity. Due to volatility, the isolated yield of 2u was moderate.

In comparison to literature results,⁵ which are shown in red in Table 2, this asymmetric Cu(I) catalysis displays marked improvement in asymmetric induction. The difference is particularly significant in the cases of the β,γ-disubstituted α,β-butenolides 2p and 2r, where the ee values were improved from 81% to 99% and from 82% to 97%, respectively. Moreover, the reaction conditions were substantially improved, i.e., rt and 0.5 h over −20 °C and 24 h.

The ambient NMR spectra of [(S)-L2Cu(MeCN)]⁺ SbF₆⁻ in CD₂Cl₂ revealed a ~ 1:1 mixture of atropisomers, which is caused by the retarded rotation of its biphenyl axis. In the atropisomer with the axis configuration identical to those in the X-ray structure in Figure 1, the ligand nitrogen lone pair electrons point away from the metal center or are shielded by the α-cyclohexyl group. As such, it is not catalytically active. We determined by NMR that the coalescence of the atropisomeric chemical shifts of this Cu(I) complex occurred between 60 °C and 65 °C and the rotational barrier was calculated to be 17.8 – 18.1 kcal/mol. At ambient temperature, the biphenyl axis is sufficiently fluxional and most of the catalyst should participate in the catalysis by adopting the desired axis configuration. We also examined the correlation between the ee of 2a and the ee of [L2Cu(MeCN)]⁺ PF₆⁻. A moderate negative non-linear effect was revealed (see SI).¹⁴ This phenomenon can be interpreted as the monomeric Cu(I) complex being the active catalyst and equilibrating homochiral dimeric/polymeric Cu(I) species such as {[(S)-L2Cu]_2•H₂O}²⁺ (PF₆ ⁻)₂ being catalytically incompetent.¹⁵

To offer insight into the reaction mechanism and understand the extraordinary asymmetric induction, we conducted DFT studies of the reaction forming 2b at the PBE1PBE level using the effective core potential SDD for Cu and the basis set 6–311g(d,p) for P and 6–31g(d,p) for the other atoms. The SMD model is employed for solvent DCM. As shown in Scheme 2, the deprotonation step eliminates the α-chiral center of the β,γ-butenolide 1b and exhibits only a minor difference in reaction barriers. The dihedral angles of Cu-O1-Cα-H in the transition states for the (S)- and (R)-1b substrates, a measure of the relative orientation of the ‘push’ and ‘pull’ in this soft enolization, are 46.9° and 72.5°, respectively, revealing deviation from orthogonality but supporting the cooperative nature of the metal and the ligand amino group in the deprotonation process. The formed (furan-2-yloxy)copper(I) intermediates 3-L2Cu and 3-L2Cu’ are conformers and differ little in free energy. The subsequent γ-protonation generates the butenolide γ-chiral center and the two TS differ in free energy by 7.7 kcal/mol, which is consistent with the observed excellent ee (i.e., 97%). In the favored TS structure TS-(R)-2b-L2Cu leading to the observed (R)-2b, the dihedral angle of Cu-O1-Cγ-H is 67.6°, while that for the disfavored TS is 1.1°. This stark difference in the relative orientation of Cu-O1 and Cγ-H reveals the former achieving substantially better metal-ligand cooperation and is attributed to the difference in reaction energy barriers. Additionally, the energy barrier of the preferred protonation is lower than that of deprotonating either of the 1b enantiomers by ≥2.3 kcal/mol, suggesting that the stereo-eliminating deprotonation is the rate-limiting step, which is opposite to that revealed by the DFT studies of the Deng’s chemistry.¹⁶

Scheme 2.

DFT Calculated energetics of the reaction forming 2b at the PBE1PBE/6–31(d,p)/6–311g(d,p)(P)/SDD(Cu) level of theory with SMD (DCM).

In summary, we have developed a rare Cu(I)-ligand cooperative catalysis that is enabled by a chiral bifunctional biphenyl-2-ylphosphine ligand. The reaction converts three-types of β, γ-butenolides into chiral α, β-butenolides with ≥96% ee. DFT calculations support the cooperative nature between the Cu(I) center and the ligand remote amino group both in the soft deprotonation and the asymmetric γ-protonation steps. Remarkably, other coinage metals, i.e., Ag and Au, can be equally effective in this catalysis. We anticipate that this coinage metal-ligand cooperation approach would find broader applications in asymmetric protonation and enolate chemistry.