Stereoconvergent Conjugate Addition of Arylboronic Acids to α-Angelica Lactone Derivatives: Synthesis of Stereochemically Complex γ-Butyrolactones

Abstract

Catalyzed stereoconvergent 1,4-additions to unsaturated carbonyls are rare but of high potential value. This letter details the development of enantioselective arylation reactions of boronic acids and β,γ-butenolides. These reactions are catalyzed by commercially available hydroxy[(S)-BINAP]-rhodium(I) dimer to afford stereochemically complex γ-butyrolactone derivatives. The reaction products provide functionality amenable to further manipulation and can lead to products with up to three contiguous stereocenters. The reaction proceeds under a dynamic kinetic resolution manifold by isomerizing the achiral starting material into an interconverting mixture of enantiomeric conjugate acceptors, followed by catalyst-controlled, enantiomer-selective 1,4-addition. Base-promoted racemization of the intermediate α,β-butenolide is possible due to the high kinetic and thermodynamic acidity of the γ-proton.

Graphical Abstract

The need for enantioenriched drug molecules has provided a significant impetus for breakthroughs in the field of dynamic kinetic resolution (DKR). DKR has proven to be a powerful tool for the synthesis of enantiomerically pure products from racemic starting materials.¹ DKR reactions commonly take advantage of a stereogenic center adjacent to an activating group to promote racemization. For example, DKR transformations of β-keto esters and α-keto esters have been widely explored, as these types of substrates can be easily racemized via the achiral enol or enolate, followed by catalyst-controlled enantio- and diastereoselective 1,2-addition (Scheme 1A).^2,3 One could imagine a complementary scenario, in which an α,β-unsaturated substrate containing a stereocenter at the γ-position could be racemized via the dienol or dienolate (Scheme 1B). One enantiomer of the starting material could then be trapped via an asymmetric 1,4-addition, to afford useful products with vicinal stereocenters at the β- and γ-positions. This reaction manifold remains unknown to the best of our knowledge, despite its direct analogy to the vast body of enantioconvergent 1,2-additions.⁴

Scheme 1. Enantioconvergent Additions of π-Functional Groups.

It may be worth considering possible reasons for this gulf, especially in the context of attempting to identify an electrophile that could potentially be appropriate for such a transformation. Successful DKR reactions of this type would require the γ-proton of the substrate to be of sufficient kinetic acidity to racemize at a rate faster than the irreversible π-bond addition. We were interested in testing the notion that α-angelica lactone (A) and related structures might serve as useful proelectrophiles for the transformation outlined in Scheme 1B. Enantioconvergent conjugate additions to α-angelica lactone are especially enticing because of that reagent’s lignocellulosic origin⁵ and the prevalence of the core scaffold in bioactive structures: nearly 10% of all natural products contain a γ-butyrolactone.⁶

Catalyzed enantioconvergent reactions are complex because of the need to engineer both a competent racemization pathway and an enantiomer-selective addition to a prochiral functional group. Scheme 2 illustrates the higher level of potential complication associated with the projected application. The desired path is A → C traced by the background gray arrow, while other mechanistically validated paths radiate from the center. The substrate A is a proelectrophile: the olefin must migrate into conjugation for it to be armed in its active form as a conjugate acceptor (A ⇆ B (≡ β-angelica lactone)). Prior reports indicate γ-substituted α,β-unsaturated butenolides racemize (B ⇆ ent-B) at reaction rates and under conditions that vary widely.⁷ Enantioenriched β,γ-substituted γ-butyrolactones are commonly accessed via diastereoselective 1,4-additions to enantioenriched γ-substituted α,β-unsaturated butenolide, allaying some concern about diastereocontrol (epi-C); however, there are no reports of accessing enantioenriched products from racemic starting materials.^6a,7d,e,8 In its “unarmed” form, α-angelica lactone is known to hydrolyze to levulinic acid (A → LA)⁹ and polymerize (nA → poly(A)),¹⁰ while the conjugated isomer B dimerizes (2 B → B-B).¹¹ Critically, identification of an appropriate catalyst that can differentiate between the two enantiomers of starting material is an inherent challenge of any stereoconvergent transformation (A → C vs A → ent-C).

Scheme 2. Kinetic Issues Involved in the Stereoconvergent Alkylation of α-Angelica Lactone (Gray Arrow).

Because the asymmetric rhodium-catalyzed 1,4-addition of arylboronic acids to butenolide substrates is well-precedented,^6a and requires conditions known to be tolerant of amine bases,^3b we believed this system would be appropriate for an enantioconvergent conjugate addition to α-angelica lactone. Herein, we report the stereoconvergent arylation of γ-substituted α,β-unsaturated butenolides via isomerization of α-angelica lactone derivatives and concomitant rhodium-catalyzed 1,4-addition of arylboronic acids.

We began our studies with reaction of lactone 1a, phenylboronic acid, and a bicyclo[2.2.2]octadiene rhodium dimer D.¹² While the reaction proceeded with high conversion and diastereoselectivity to the desired product 3a, the reaction displayed only modest enantioselectivity and was accompanied by an appreciable amount of the undesired dimer B-B (Table 1, entry 1). By switching the aryl group on the complex E, dimerization could be greatly suppressed; however, both diastereoselectivity and enantioselectivity of the transformation were compromised (Table 1, entry 2). Employing the commercially available (S)-BINAP hydroxy-rhodium dimer catalyst F,¹³ we obtained poor conversion to the desired product 3a but observed excellent enantio- and diastereoselectivity (Table 1, entry 3). Further investigation revealed that CsF rapidly induces dimerization of the α-angelica lactone, and when it was omitted, dimerization did not occur (Table 1, entry 4). Given the inorganic base’s role in facilitating transmetalation, we considered alternative additives in order to increase conversion.¹⁴ The addition of K₂CO₃ successfully improved conversion, but also accelerated the undesired hydrolysis of α-angelica lactone to levulinic acid, LA (Table 1, entry 5). Water is necessary to achieve high conversion to the desired product (Table 1, entry 6). Additionally K₂CO₃ decreases the enantioselectivity of the reaction, a fact that we attribute to a relatively fast rate of arylation relative to the rate of racemization of the substrate. Our group previously reported that the rate of racemization of configurationally labile substrates in the context of DKR reactions can be accelerated by smaller organic bases.^3b Accordingly, use of N-methylpyrrolidine (NMP), a less sterically encumbered organic base, restored excellent levels of enantioselectivity (Table 1, entry 7). Although hydrolysis is suppressed at lower temperatures, arylation is slowed significantly, leading to increased dimerization (Table 1, entry 8). Increasing the amount of organic base had a dual effect of accelerating isomerization to the requisite butenolide isomer, effectively inhibiting hydrolysis, and promoting faster racemization relative to the arylation to afford the desired product in good enantioselectivity (Table 1, entry 9). By use of 10 equiv of N-methylpyrrolidine, the desired product was prepared in excellent conversion, diastereo- and enantioselectivity in 1 h (Table 1, entry 11). Additional studies confirmed that dioxane, K₂CO₃, and arylboronic acid are the best solvent, inorganic base, and organoboron source, respectively (see Supporting Information).

Table 1.

Reaction Optimization

entrya	catal. (mol %)	org. base (equiv)	inorg. base (equiv)	conversion, % (3a/B-B/LA)	erb
1c,d	D (2.5)	Et3N (6)	CsF (6)	100 (1.7:1:0)	24:76
2c,d	E (2.5)	Et3N (6)	CsF (6)	100 (11:1:0)	44:56
3e	F (2.0)	Et3N (1)	CsF (2)	15 (1.5:1:0)	92:8
4	F (2.0)	Et3N (1)	none	7 (1:0:0)	97:3
5	F (2.0)	Et3N (1)	K2CO3 (1)	100 (2.5:0:1)	85:15
6f	F (2.0)	Et3N (1)	K2CO3 (1)	20 (1:0:0)	i
7	F (2.0)	NMpg (1)	K2CO3 (1)	100 (6.7:0:1)	96:4
8e	F (2.0)	NMPg (1)	K2CO3 (1)	56 (3:1:0)	97:3
9	F (2.0)	NMPg (6)	K2CO3 (1)	80 (5.0:0:1)	96:4
10	F (2.0)	NMPg (10)	K2CO3 (1)	100 (10:0:1)	95:5
11h	F (2.0)	NMPg (10)	K2CO3 (1)	90 (1:0:0)	95:5

^aAll reactions were run on a 0.20 mmol scale.

^bDetermined by HPLC using a chiral stationary phase.

^cReaction solvent CH₂Cl₂.

^dReaction run at room temperature.

^eReaction T = 40 °C.

^fReaction solvent 1,4-dioxane.

^gNMP = N-methylpyrrolidine.

^hReaction time = 1 h.

ⁱNot determined.

With optimized conditions in hand, we sought to explore their applicability to other reaction partners, beginning with unsaturated lactones bearing other substituents at the γ-position (Table 2). Substrates containing both short and long alkyl chains afforded the desired products, 3a, 3b, and 3c, in good yields and excellent stereoselectivities. Homologated carbocycles at the γ-position gave desired product 3d in good yield and enantioselectivity. The benzyl group is also tolerated at the γ-position and gives lactone 3e in excellent yield and stereoselectivity. A branched alkyl group worked well in the transformation to afford 3f, albeit in slightly diminished yield, presumably due to the steric bulk at the γ-position. All products were obtained as a single diastereomer.

Table 2.

Scope of γ-Substituted β,γ-Unsaturated Butenolides

^aExcess boronic acid is needed because of background hydrolysis (to benzene, see ref 16) and competitive angelica lactone dimerization.

The scope of boronic acids tolerated in this reaction was then explored (Table 3). We found a wide variety of para-substituted arylboronic acids could be employed to generate products containing electron-donating groups, such as p-tolyl 3g and p-anisyl 3h. We could also obtain products containing para-substituted halogens, 3i–k, and electron-withdrawing para-ester 3l in modest to good yields and good enantioselectivities. Electron-rich arylboronic acids highlight the delicate balance between the rates of arylation and racemization. Reactions employing electron-rich arylboronic acids require catalytic quantities of K₂CO₃ to achieve desirable levels of enantioselectivity; a full equivalent of inorganic base results in loss of enantioselectivity. We hypothesize that the erosion is due to the accelerated arylation relative to racemization. This trend continues for meta-substituted arylboronic acids. Products with electron-releasing substituents at the meta-position, 3m and 3n, can be obtained in good yields and enantioselectivities when substoichiometric K₂CO₃ is used. Halogen substitution at the meta-position (3o) is tolerated as well. Pleasingly, a sterically bulky ortho-substituted arylboronic acid provided product 3p with excellent enantioselectivity, though greatly decreased yield. Naphthylboronic acid coupled product 3q can be attained. Electron-rich heteroaromatic boronic acids work to afford heteroaromaticcontaining lactones 3r and 3s. X-ray diffraction studies revealed the stereochemistry of 3g, 3j, and 3q to be (4S,5S) (See Supporting Information).¹⁵ The absolute stereochemistry agrees with the results of Hayashi and co-workers.^13,14,16,17 On a 1 mmol scale using 1 mol % catalyst C, α-angelica lactone and phenylboronic acid reacted to provide the derived lactone 3a in 89% yield, >20:1 diastereoselection, and 93:7 er.

Table 3.

Scope of Arylboronic Acid^b

^aReaction run with 5 mol % K₂CO₃

^bResults in brackets are for reactions run with 4.0 equiv of ArB(OH)₂.

Current limitations of the substrate scope include electronpoor heteroaromatic and styrenyl boronic acids.^3b Higher yields can generally be achieved by employing more arylboronic acid, but in cases of electron-rich arylboronic acids, enantioselectivity typically suffers, presumably due to the rate of arylation relative to the rate of racemization. We hypothesize that under the optimized conditions, protodeborylation of the arylboronic acids occurs rapidly, accounting for moderate yields with some boronic acids. In most cases, longer reaction times do not improve yields.¹⁴

Further chemical transformations of the enantioenriched β,γ-substituted γ-butyrolactones were performed in order to access synthetically useful organic intermediates (Scheme 3). The disubstituted lactone 3a can be reduced to 1,4-diol 4a with a primary and secondary alcohol. Lactone amidolysis of 3a with ammonium hydroxide afforded enantioenriched amide 5a, which can be further reduced to afford its derived 1,4-amino alcohol (see the Supporting Information). Finally, the β,γ-substituted γ-butyrolactone 3a can be alkylated at the α-position in excellent diastereoselectivity to afford γ-butyrolactone 6a with three contiguous stereocenters. The relative stereochemistry of this transformation was confirmed by a 1D NOE NMR study (see the Supporting Information).¹⁸

Scheme 3. Synthetic Utility of Lactone Products^a.

^aReagents and conditions: (a) LiAlH₄, THF, 23 °C, 2 h; (b) aq. NH₄OH, 23 °C, 5 h; (c) LDA, allyl iodide, THF, −78 to 0 °C, 3 h.

When the catalyzed addition of 2l to 1a was carried out in dioxane/D₂O (10:1), partial deuterium incorporation was observed at both the α- and γ-positions of 3l (Scheme 3B). The fact that the product is obtained with less than 100% of ¹H/²H exchange carries ramifications for the mechanism of the obligatory 1,3-prototropy. The results in part implicate [R₃N–H]⁺, rather than D₂O, as the partner that reacts with the dienolate^7b and concurrently confirm the importance of the organic base in the process.

In summary, we have developed an enantioconvergent 1,4-arylation of γ-substituted α,β-unsaturated butenolides. This reaction is one of the first dynamic kinetic resolutions that takes advantage of racemization of a γ-stereocenter through formation of a dienolate. A wide range of β-aryl, γ-substituted γ-butyrolactones can be accessed in good yields, excellent enantioselectivities, and as a single diastereomer. These lactones can be further manipulated to generate useful organic building blocks, including 1,4-diols and 1,4-amino alcohols. Additionally, the products can be diastereoselectively alkylated to afford stereochemically complex trisubstituted lactones. Extension of this work to other classes of nucleophiles, as well as application of the DKR conjugate addition manifold to additional substrates is currently underway in our laboratory.