Enantioselective and Diastereodivergent Allylation of Propargylic C–H Bonds

Abstract

An iridium-catalyzed stereoselective coupling of allylic ethers and alkynes to generate 3,4-substituted 1,5-enynes is reported. Under optimized conditions, the coupling products are formed with excellent regio-, diastereo-, and enantioselectivities, and the protocol is functional group tolerant. Moreover, we report conditions that allow the reaction to proceed with complete reversal of diastereoselectivity. Mechanistic studies are consistent with an unprecedented dual role for the iridium catalyst, enabling the propargylic deprotonation of the alkyne through π-coordination, as well as the generation of a π-allyl species from the allylic ether starting material.

Graphical Abstract

The enantioselective functionalization of C(sp³)–H bonds constitutes a conceptually simple and direct approach for the synthesis of valuable stereodefined products from simple starting materials.¹ Given the widespread availability of alkyne derivatives,² the enantioselective construction of a C–C or C–X (X = N, O, etc.) bond at the propargylic position is an especially attractive strategy for accessing versatile chiral building blocks en route to stereochemically and functionally complex targets.^3,4 However, in contrast to a variety of well-developed protocols that employ prefunctionalized alkyne derivatives for asymmetric propargylation^5–7 or metal acetylide precursors for enantioselective nucleophilic addition to synthesize α-chiral alkynes, ^8–9 the use of simple alkynes for enantioselective propargylic C–H functionalization remains an underdeveloped approach (Scheme 1A).^10–12

Scheme 1.

Asymmetric propargylic functionalization for 1,5-enyne synthesis

Metal nitrene or carbene insertion constitutes one general strategy for enantioselective propargylic C–H functionalization (Scheme 1B, top). A number of transition metals and ligand systems have been reported to facilitate enantioselective nitrene (or nitrene radical) transfer,^{10a–h,11a–c} Likewise, several examples of enantioselective carbene insertion into a propargylic C–H bond have been disclosed.^10i,11d In addition, enantioselective propargylic C–H functionalization proceeding via radical–organometallic crossover mechanisms are also known, constituting another promising strategy. ¹² Given their potential applicability toward establishing stereochemistry at both the α- and β-positions of the alkyne,^5f,8i,11d the development of new stereoselective propargylic C–H functionalization strategies has continued to attract interest.

Our research group and others have previously demonstrated an alternative approach to catalytic propargylic C–H functionalization wherein deprotonation of an α-C–H bond of a metal-bound alkyne enables the subsequent electrophilic trapping of the resultant allenylmetal nucleophile to deliver the product of propargylic functionalization (Scheme 1B, bottom).^13,14 Zhang and coworkers have further demonstrated that such a process could be rendered enantioselective in the intramolecular sense using a chiral Au catalyst.^14a

In this context, we wondered whether we could develop an intermolecular coupling strategy by intercepting the allenylmetal species with a stereodefined organometallic electrophile.¹⁵ Although originally envisaged as a cooperative catalysis strategy,¹⁶ we disclose herein our serendipitous finding that a single dual-function catalyst enables the simultaneous activation and subsequent coupling of two starting materials,¹⁷ an alkyne and an allylic ether, to generate synthetically versatile 1,5-enyne products with high levels of regio– and stereoselectivity (Scheme 1C).¹⁸

We initially hypothesized that, by combining our previously reported Fe-based catalysts for propargylic C–H deprotonation¹³ with an Ir-based system for enantioselective allylic substitution,^19,20 we could develop a cooperative strategy for the enantioselective synthesis of 1,5-enynes by intercepting a chiral, non-racemic electrophilic π-allyliridium intermediate with a nucleophilic σ-allenyliron intermediate. To validate the proposed reactivity, we examined the stoichiometric reaction of an Fp*-based σ-allenyliron species (Fp* = (η⁵-C₅Me₅)Fe(CO)₂)^13a towards a previously characterized π-allyliridium complex (A⁺TfO⁻) bearing a phosphoramidite–alkene ligand system developed by Carreira and coworkers.^20a Indeed, the desired product was obtained in moderate yield and excellent enantioselectivity, albeit with little control over the diastereoselectivity (1:1.7 dr) (Scheme 2A).

Scheme 2.

Preliminary results for propargylic C–H allylation

Next, we attempted to implement a catalytic system by using 1-phenyl-1-butyne and allylic ether 2a as starting materials while employing substoichiometric amounts of Fe (20 mol %) and Ir (3 mol % [Ir(cod)Cl]₂, 6 mol % based on Ir) complexes. In addition, in analogy to our previously reported Fe-catalyzed α-C–H functionalization reactions, BF₃·OEt₂ and 2,2,6,6-tetramethylpiperidine (TMPH) were included in the reaction mixture to promote ionization of the allylic ether and to deprotonate the metal-alkyne complex, respectively.¹³ These conditions afforded the same coupling product, though with lower yield and significant differences in the stereoisomeric composition (Scheme 2B). During the course of optimization, we performed a series of control experiments and found, to our surprise, that the reaction proceeded in excellent yield without the need for the iron complex.²¹ These results suggested the possibility that, in addition to facilitating the generation of the π-allyl electrophile, the Ir complex could serve the additional role of activating the alkyne to allow for the removal of a propargylic proton under mildly basic conditions. To the best of our knowledge, this second role has not previously been demonstrated for Ir–phosphoramidite complexes.^20k

After extensive optimization of the identity and stoichiometry of allylic electrophile, base, ligand and Lewis acid, conditions that delivered the anti-diastereomer 3a in high yield and superb stereoselectivity (91% isolated yield, >20:1 dr, >99% ee) were identified (Table 1, entry 1). Of note, the use of additional allylic methyl ether (3.0 equiv) was found to improve diastereoselectivity (entry 1 vs. 2). Moreover, the ligand-to-metal ratio was found to be critical (entry 1 vs. 7).^8i,20b,20j Surprisingly, when achiral ligand L₄ was employed, a complete reversal of the diastereoselectivity was observed, with syn-diastereomer 4a formed in >20:1 dr though in low yield (entry 9). By using racemic ligand (±)-L₁, 4a could be obtained in excellent yield and diastereoselectivity (entry 10). Control experiments showed that the reaction fails in the absence of [Ir(cod)Cl]₂, (S)-L₁, TMPH or BF₃·OEt₂, indicating the necessity of each component (entry 13).

Table 1.

Optimization of the Ir-catalyzed alkyne-allyl coupling reaction^a

entry	variation	% yield (% ee)	3a:4ab
1	none	92c (>99)	>20:1
2	2a (2.0 equiv)	88 (>99)	5.1:1
3	2ab instead of 2a	NPd	--
4	2ac instead of 2a	69 (>99)	>20:1
5	Et3N, DBU, or KOtBu as base	NP	--
6	BPh3 or B(C6F5)3 as Lewis acid	NP	--
7e	[Ir]:(S)-L1 = 1:1, with 3 mol % Ir	NP	--
8e	(S)-L2 or (R)-L3 as ligand	NP	--
9e	L4 as ligand	<10	<1:20
10f	(±)-L1 as ligand	99	<1:20
11f	[Ir]:(S)-L1 = 1:2, with 1 mol % Ir	19 (>99)	>20:1
12	at rt (23 °C)	70 (>99)	>20:1
13	no [Ir], (S)-L1, TMPH, or BF3·OEt2	NP	--

^aOn 0.1 mmol scale. Yields (3a+4a) were determined by ¹H NMR spectroscopy of the crude reaction mixture, using 2,4-dinitrotoluene as the internal standard. Enantiomeric excesses were determined by chiral HPLC.

^bDiastereomeric ratios (3a:4a) were determined by ¹H NMR spectroscopy of the crude mixture.

^cIn 91% isolated yield.

^dNP: no desired product observed.

^eCH₂Cl₂ (1 M) as solvent.

^f2.0 equiv 2a was added.

With optimized conditions in hand, we investigated the generality of this propargyl–allyl coupling protocol with respect to the alkyne coupling partner (Table 2). A collection of aryl ethyl acetylenes bearing electron-donating (3b, 3c, 3g, 3h) or electron-withdrawing substituents (3d, 3e) were transformed into the corresponding 1,5-enynes in moderate to high yields and excellent diastereo- and enantioselectivities. Furthermore, a number of heterocycles were accommodated, including a thiophene (3i), a furan (3j) and a benzofuran (3k). An alkyne bearing a phthalimido group (3l) also reacted efficiently with high stereoselectivity. Moreover, higher aryl alkyl acetylenes including those possessing pendent ether or sulfonate ester groups (3m–3o) were also competent substrates. The synthesis of 3n bearing a p-toluenesulfonate ester was carried out on 5 mmol scale to deliver 1.58 g of the product (73% isolated yield, >99:1 dr, >99% ee) after recrystallization of the crude material. A conjugated enyne (3p) likewise delivered a successful outcome. The primary propargylic C–H bonds of aryl methyl acetylenes could also be functionalized (3qa–3qb),^11a,22 giving the final product with good enantioselectivity, albeit in only moderate yields. Finally, we tested the reactivity of a dialkyl acetylene, 2-hexyne (3r). Only one regioisomer was isolated in excellent enantiomeric excess, though the low yield demonstrates a limitation of our current protocol.

Table 2.

Substrate scope for the alkyne coupling partner^a

^aIsolated yields. Enantiomeric excesses were determined by chiral HPLC. Diastereomeric ratios were determined by ¹H NMR spectroscopy of the crude material.

^bCH₂Cl₂ (0.2 mL) as solvent.

^c[Ir(cod)Cl]₂ (6 mol % Ir), (S)-L₁ (12 mol %) were used.

^d[Ir(cod)Cl]₂ (4 mol % Ir), (S)-L₁ (8 mol %), 36 h; purified by recrystallization.

Subsequently, the scope with respect to the allylic ether coupling partner was examined (Table 3). A range of racemic allylic ethers could be coupled to give the desired 1,5-enynes products in moderate to high yields as a single diastereomer. Aryl groups with various substitution patterns (5a–5i) were compatible substrates, as was the presence of a trialkylsilyl substituent (5j), a sulfonate ester (5k), a sulfonamide (5l), a carboxylic ester (5m), and a tertiary amide (5n).

Table 3.

Substrate scope for the allylic ether coupling partner^a

^aIsolated yields. Enantiomeric excesses were determined by chiral HPLC. Diastereomeric ratios were determined by ¹H NMR spectroscopy of the crude material.

^bCH₂Cl₂ (0.2 mL) as solvent.

We then examined a subset of the alkyne/allylic ether combinations from Tables 2 and 3 to probe the generality of the reaction conditions employing (±)-L₁ that selectively affords the syndiastereomer of the 1,5-enyne (Table 4). In all cases examined, nearly complete reversal of the reaction diastereoselectivity was observed. Moreover, the yields of coupling products (±)-4 and (±)-6 were generally comparable or higher than those observed for the formation of the corresponding diastereomeric products 3 and 5. Notably, the synthesis of (±)-4a could be performed on 1 mmol scale with near quantitative yield, even at a reduced catalyst loading of 2 mol % Ir.

Table 4.

Diastereoselective synthesis of the syn-isomer^a

^aIsolated yields. Diastereomeric ratios were determined by ¹H NMR spectroscopy of the crude material.

^b[Ir(cod)Cl]₂ (2 mol % Ir), (±)-L₁ (4 mol %) were used, 24 h.

^cCH₂Cl₂ (0.2 mL) as solvent.

The synthetic value of this method was further demonstrated through the preparation of diverse carbo– and heterocyclic scaffolds with high molecular complexity (Scheme 3). The Aucatalyzed cycloisomerization of 1,5-enynes efficiently afforded bicyclo[3.1.0]hexene 7n in 91% yield with >99% ee and 12:1 dr.^18a Moreover, tricyclic indole derivative 8n was synthesized via a palladium-catalyzed cascade process in 52% yield without the loss of enantioselectivity.²³ Finally, a regio- and stereoselective hydroboration–oxidation sequence, followed by O-tosylation and cyclization with BnNH₂ as a lynchpin, afforded 4,5-disubstituted azepane 9n in good yield and excellent enantioselectivity.

Scheme 3. Diversification of product 3n^a.

^aConditions: (a) Ph₃PAuCl (cat.), AgSbF₆ (cat.), CH₂Cl₂, rt, 0.5 h; (b) Sulfonamide 8S, Pd(OAc)₂ (cat.), Ph₃P (cat.), Cs₂CO₃, CsI, norbornene, DMF, 130 °C, 12 h; (c) 9-BBN, THF, 2h; NaBO₃·H₂O, H₂O, rt, 3 h; (d) TsCl, Et₃N, DMAP, CH₂Cl₂, rt, 4 h; (e) BnNH₂, K₂CO₃, MeCN, reflux, 22 h.

We performed a series of additional experiments to gain some insight into the mechanistic details of the catalytic system (Scheme 4). First, a plot of conversion of 1a over time revealed that the reaction was subject to an induction period. While this complicated efforts to measure rate constants, we next subjected ethyl-deuterated 1a-d₅ to the same conditions and observed a smaller maximum velocity and slower overall rate, indicating a significant kinetic isotope effect (KIE). Under conditions of intermolecular competition, a KIE of k_H/k_D = 4.5 was measured (Scheme 4A). Combined, these data suggest rate-limiting proton abstraction at the propargylic position.²⁴

Scheme 4.

Selected mechanistic studies and proposed catalytic cycle

To probe the nature of the stereocontrol of this transformation, we investigated the relationship between the enantiomeric excess of the catalyst and the diastereo- and enantiomeric composition of the product (Scheme 4B). The observation of non-linear effects²⁵ (particularly for formation of product 3a) suggests the presence of two ligands on the metal center in the enantiodetermining transition states. Moreover, the (S,S)-(L₁)₂[Ir] homochiral complex is primarily responsible for formation of 3a, while formation of 3a from the (R,S)-(L₁)₂[Ir] heterochiral complex is slow. The effect of the enantiomeric composition of 2a was then examined (Scheme 4C). When racemic 2a was used, a kinetic resolution was observed, with the less reactive (R)-2a accumulating in the reaction mixture. In addition, a large difference in the diastereomeric ratio of the product was observed when (S)- and (R)-2a were used as starting materials.²⁶ This suggests that departure of the methoxy group of the iridium-coordinated allylic ether takes place after coordination (and perhaps deprotonation) of the alkyne coupling partner.^20d

On the basis of these results, a plausible though still speculative mechanism is proposed in Scheme 4D. Carreira’s complex (in the form of A⁺[BF₃OMe]⁻) was observed to be the sole Ir–phosphoramidite species by ³¹P{¹H} NMR analysis of the reaction mixture at the beginning of the reaction and continued to be the major phosphorus-containing species in the reaction mixture up to 14 h later.^20a Moreover, A⁺TfO⁻ was found to be a competent catalyst for the reaction (89% NMR yield, >20:1 dr, >99% ee). Thus, we postulate that 18-electron complex A⁺ acts as an initial reservoir for on-cycle catalytically active species. In particular, we propose that formation of allylic ether complex I with loss of Cl⁻ allows for binding of the alkyne,²⁷ with the key deprotonation step taking place from cationic intermediate II.²⁸ Subsequent abstraction of methoxide from the bound allylic ether and ligand coupling with reduction at Ir then furnishes 3a.^6a,29 Finally, coordination of the allylic ether to V regenerates I to close the catalytic cycle.

In summary, we have described the Ir-catalyzed enantioselective and diastereodivergent propargylic C(sp³)–H allylation of alkynes, including a mechanistically novel role for the catalyst. Further studies of the detailed mechanism and explorations of additional applications of this strategy are ongoing and will be reported in due course.