Atroposelective interrupted CuAAC reaction using cyclic diaryliodoniums

Abstract

Copper-catalyzed azide–alkyne cycloaddition (CuAAC) is a pivotal strategy for joining two fragments, including bioactive moieties under mild conditions. In recent years, “interrupted” CuAAC reaction has emerged by intercepting copper triazolide intermediates, thereby significantly broadening the scope of these transformations. However, asymmetric transformations based on the interruption of copper triazolide intermediates have remained elusive. In this work, we present an efficient and mild approach that intercepts a copper triazolide intermediate with a cyclic diaryliodonium reagent, enabling a three-component coupling of the cyclic diaryliodonium species, an alkyne, and an azide under atroposelective control. This method constitutes an asymmetric interrupted CuAAC reaction and furnishes a diverse array of structurally unique atropisomeric biaryl triazoles. Mechanistic investigations reveal an unusual secondary kinetic isotope effect for the terminal alkyne, while in situ calorimetry–based reaction progress kinetic analysis identifies the individual reaction orders of each substrate: formal 0.4th-order for cyclic diaryliodonium and first-order for both the alkyne and the azide. The overall rate is governed by copper-mediated cyclometallation of the alkyne and azide, as well as the oxidative addition of the copper triazolide intermediate to the cyclic diaryliodonium reagent.

Introduction

Since the independent development of copper-catalyzed azide-alkyne cycloaddition (CuAAC) reaction by Sharpless¹ and Meldal² groups, CuAAC reaction has been widely used in organic synthesis, materials science, bioconjugation and chemical biology (Fig. 1a)^{3,4,5,6,7,8,9}. The discovery of copper as a catalyst offers two key advantages: (a) The copper catalyzed reaction yields highly regioselective 1,4-triazoles, and lowers the reaction temperature compared to the classic Huisgen cycloaddition between azide and alkyne (e.g., reflux in toluene), thereby enabling applications in biochemistry; (b) The mechanistic studies have confirmed the existence of a key metallocycle intermediate¹⁰, the 1,4-substituted triazolyl copper species M1, which have ever been unambiguously characterized by X-ray single crystal analysis^11,12. Building on these findings, a range of interrupted reactions have been developed (Fig. 1a). After the pioneer work of Wu et al.¹³, a variety of electrophiles, such as allyl halide¹⁴, RS(O)₂SR¹⁵, acid chloride¹⁶, CuI/N-bromosuccinimide¹⁷, SS-tert-butyl p-toluenesulfono(dithioperoxoate)¹⁸, Bu₃SnOMe¹⁹, diselane²⁰, and nitrone²¹ etc.^{22,23,24,25,26}, can be used to terminate the triazolyl copper species, a process typically referred to as the interrupted CuAAC reaction. Alternatively, less reactive species, i.e., sp²-hydridized aryl bromides and iodides^27,28,29,30, can serve as terminating agents when utilized as intramolecular components to access cyclic 5-aryl-1,4-triazoles.

Fig. 1: CuAAC reaction and Cu-catalyzed coupling of cyclic diaryliodoniums.

a Interrupted CuAAC reaction. b Cu-catalyzed asymmetric ring-opening reaction of cyclic diaryliodoniums. c Our design plan: atroposelective interrupted CuAAC with cyclic diaryliodoniums.

Motivated by wide applications of atropisomers (particularly axially chiral biaryl compounds)^{31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48}, we have recently developed a copper-catalyzed highly stereoselective ring-opening reaction of cyclic diaryliodoniums to construct tetra-substituted biaryl atropisomers^49,50. The retained C-I bonds in the product can be used for late-stage diversification⁵¹. This reaction is believed proceed through an oxidative addition of Cu(I) with cyclic diaryliodoniums to form Cu(III) intermediate M2, followed by reductive elimination to achieve the formation of carbon-heteroatom bonds (Fig. 1b). However, to date, apart from palladium-catalyzed carbonylation reactions by Hayashi⁵², Wu and Liao⁵³, only heteroatom nucleophiles have been successfully applicable to this Cu-catalyzed asymmetric ring-opening reaction⁵⁴.

In this work, we combine these two catalytic cycles, Cu(I)-Cu(I) cycle and Cu(I)-Cu(III) cycle, to realize an enantioselective diaryliodonium-interrupted CuAAC reaction, allowing for the participation of a formal carbon nucleophile in Cu-catalyzed ring-opening of cyclic diaryliodoniums (Fig. 1c). In addition to the significant challenge of controlling the enantioselectivity, careful management of competing side-reaction pathways is essential. For instance, three Cu(I) species, L*Cu(I), RC≡CCu(I), triazolyl copper(I) M1, might compete in the oxidative addition reaction with cyclic diaryliodonium, leading to the formation of three distinct Cu(III) intermediates. We hypothesize that the sp²-hybridized carbon-copper M1 is more electron-rich than either L*Cu(I) or the sp-hybridized carbon copper RC≡CCu(I), suggesting that M1 may exhibit a higher oxidative tendency toward diaryliodoniums than the other two copper species. Moreover, in aprotic solvent, the final protonation of CuAAC intermediate M1 is usually a slow process, and generally regarded as the rate-determining step⁵⁵, providing an opportunity to trap this intermediate by diaryliodoniums, rather than through a slow protonation with terminal alkynes.

Results and discussion

Optimize the reaction conditions

In light of these considerations, we first selected functionalized alkyne 2a and azide 3a as substrates. Copper(II) triflate [Cu(OTf)₂] was ineffective, despite the usual expectation that Cu(II) can undergo disproportionation to generate Cu(I) and Cu(III) or reduction to Cu(I) species via alkyne homocoupling (Table 1, entry 1)⁵⁶. In contrast, the Cu(MeCN)₄PF₆/L1 catalytic system successfully catalyzed the reaction, giving a moderate yield of 4a with good enantioselectivity (entry 2). Product 4a contains two potential chiral axes. Nuclear magnetic resonance (¹H NMR) analysis shows a diastereomeric ratio (dr) of ~15:1, which varies with deuterium solvent choice (e.g., CDCl₃, CD₂Cl₂, DMSO-d₆, toluene-d₈; see the Supplementary Fig. S1). These observations indicate that the phenyl-phenyl axis is stable, whereas the phenyl–triazole axis is semi-stable, undergoing rotation at room temperature. Reducing the reaction concentration improved the yield but slightly lowered enantioselectivity (entry 3). Increasing the proportion of 2a and 3a relative to 1a provided no further yield improvement and instead diminished enantioselectivity (entry 4). Changing bases and modifying the electronic nature of the pyridinyl substituents in pyridine-2,6-bis(oxazoline) PyBox (L2–L4) did not provide better results (entries 5–9). We subsequently varied the amino-alcohol backbone and its electronic properties (L5-L9), but none yielded improved outcomes (entries 10–14). Notably, bis(oxazoline) ligand L10 showed worse performance than pyridine-2,6-bis(oxazoline) ligand L1 (compare entry 15 with entry 3). In addition, using K₂CO₃ flame-dried under vacuum in a Schlenk tube further improved the reaction and exhibited good reproducibility (entry 16). The absence of 4 Å molecular sieves resulted in an improved yield, but a considerable drop in enantioselectivity (entry 17).

Table 1 Reaction condition optimization^a

Substrate scope

Upon identifying the optimal reaction conditions, we examined the scope of the reaction (Figs. 2 and 3). First, we focused on the influence of various substituents on arylalkynyl ketones. Overall, most para-substituents on phenylalkynyl ketones had minimal impact on yield or enantioselectivity (4a–4i), except for the nitro substituent, which led to a decreased enantioselectivity (83% ee) (4j). The structure and absolute configuration of compound 4j (CCDC 2421103, see the Supplementary Table S3) were confirmed by single-crystal X-ray diffraction study. Large meta-substituents enhanced enantioselectivity (e.g., 4o) but often reduced the yield; similarly, a 3,4-dimethoxy substituent offered a low yield of 4r. Heteroaryl derivatives, such as those bearing 3-thiophene and 3-methyl-2-benzofuryl moieties, also reacted smoothly to afford the desired products. In general, ortho-substituents diminished the efficiency of this interrupted CuAAC reaction, resulting in somewhat reduced yields (4v–4z). 1-Alkyl-2-propyn-1-one derivative showed a slight drop in enantioselectivity (4aa), while the reaction with N-tolyl propargyl amide gave an even lower enantioselectivity of 76% ee (5). To explore the reaction compatibility of different substituted alkynes and azides, we conducted automated substrate screening using a chemical robot and identified suitable candidates for this three-component cross-coupling reaction (Supplementary Table S1). When 2-(prop-2-yn-1-yl)isoindoline-1,3-dione was employed as a substrate, the reaction afforded the product (6) with good enantioselectivity, albeit with significantly diminished yield. Aryl propargyl ethers proved to be excellent substrates, delivering good enantioselectivity while tolerating various substituents on the benzene ring, including para-acetyl, para-nitro, para-trifluoromethyl, meta-bromo, and ortho-bromo groups, etc (7a-7j). Similarly, 2-naphthyl propargyl ether achieved excellent enantioselectivity, though the corresponding product was obtained in only 37% yield (7k). Propargyl benzyl ether also demonstrated good compatibility with this coupling system, furnishing 7 l in 49% yield with 98% ee. In contrast, phenylacetylene and para-toluenesulfonyl acetylene proved incompatible with this reaction, resulting in either poor enantioselectivity or low yield (8 and 9).

Fig. 2: Substrate scope of alkyne.

Conditions: 1 (0.1 mmol), 2 (0.12 mmol), 3 (0.12 mmol), Cu(MeCN)₄PF₆ (0.010 mmol), L1 (0.02 mmol), 4 Å molecular sieves (80 mg), K₂CO₃ (3.0 equiv.) in CH₂Cl₂ (4 mL), and stirred at 25 °C for 36 h. Isolated yields are reported, and ee values were determined by chiral stationary HPLC.

Fig. 3: Substrate scope of diaryliodonium and azide.

Conditions: 1 (0.1 mmol), 2 (0.12 mmol), 3 (0.12 mmol), Cu(MeCN)₄PF₆ (0.010 mmol), L1 (0.02 mmol), 4 Å molecular sieves (80 mg), K₂CO₃ (3.0 equiv.) in CH₂Cl₂ (4 mL), and stirred at 25 ^oC for 36 h. Isolated yields are reported, and ee values were determined by chiral stationary HPLC.

Regarding the azide component, the scope remains largely restricted to α-azido acetate derivatives and benzylic azides. α-Azido acetates/amides derived from methanol, benzyl alcohols, or benzylamine all furnished good yields and stereoselectivities (10a, 10c–10k), whereas the phenol-based ester exhibited substantially lower efficiency and enantioselectivity (10b). Benzylic azides bearing para-, meta- and ortho-substituents, as well as 2-(azidomethyl)naphthalene, demonstrated good reactivity with enantioselectivity ranging from 76–91% (10l-10r). Notably, when TsN₃ was employed as the substrate, no three-component cross-coupling product 11 was detected. Next, we explored the influence of various substituents on the cyclic diaryliodonium scaffold. Introducing additional methyl groups at the meta or para positions relative to the C–I bond maintained good yields and enantioselectivities (12a and 12b). However, with a few exceptions (e.g., 12c and 12e), incorporating ortho-methyl or chloro group at the C–I bond significantly reduced the yields (12d and 12f). Cyclic diaryliodoniums with fluoro groups at the meta or para positions relative to the C–I bond resulted in diminished enantioselectivity (12g and 12h). The cross-coupling reactions involving phenyl propargyl ether and α-azido acetate with binaphthenyl cycloiodonium or 1,9-dibromodibenzoiodonium gave excellent enantioselectivity, albeit with fairly low mass yields (12i and 12j). Finally, 1,9-bis(tosyloxy)dibenzo[b,d]iodonium proved ineffective as a substrate, with no desired product being detected (12k).

Mechanistic studies

We subsequently investigated the mechanism of this enantioselective interrupted CuAAC reaction. Alkynyl copper species 2a-Cu was prepared independently, which can smoothly react cyclic diaryliodonium 1a to give the alkynylation product 13 in moderate yield (Fig. 4a). However, the catalytic reaction between 1a and 2a proceeded significantly more slowly than the three-component coupling involving 1a, 2a, and 3a, resulting in the formation of 13 in 22% yield with only 36% enantioselectivity, which contrasted sharply with the high enantioselectivity observed in the three-component reaction (Fig. 4b and 4c). Furthermore, attempted cycloaddition reaction between 13 and azide 3a under the standard condition failed to proceed. These findings clearly ruled out a mechanism in which the three-component coupling product was simply aroused from a subsequent [3 + 2] cycloaddition between 13 and 3a.

Fig. 4: Mechanistic experiments.

a The reaction of alkynyl copper 2a-Cu with cyclic diaryliodonium. b Catalytic asymmetric reaction between 1a and 2a. c Catalytic reaction between 13 and 3a.

Next, we investigated the standard [3 + 2] cycloaddition reaction between 2a and 3a, which proved highly sensitive to reaction conditions: in the absence of an additional base and molecular sieves, the reaction proceeded slowly (Fig. 5a). Subsequently, reaction between cyclic diaryliodonium 1a and triazole 14 was tested under standard conditions, and no final product 4a was detected, ruling out the possibility of a Cu-catalyzed C-H activation-initiated ring-opening reaction between cyclic diaryliodonium 1a and compound 14. Notably, K₂CO₃ accelerated the reaction but a dimeric triazole 15 can be detected (Fig. 5b). Replacing alkyne 2a with the copper acetylide 2a-Cu in the three-component reaction yielded 20% of 4a (48% ee), 22% of 13, along with a small amount of 14 and the dimeric product 15 (Fig. 5c)⁵⁷. These results deviate from the those in Fig. 2, suggesting that 2a-Cu competes with the copper triazolide intermediate for oxidative addition with the cyclic diaryliodoniums, and is therefore unlikely to be the active intermediate in the catalytic version employing 1a, 2a, and 3a. Further monitoring by nuclear magnetic resonance (NMR) showed that 4 Å molecular sieves slowed the overall process, and also diminished the byproduct 14 (Fig. 5d, and Supplementary Table S2).

Fig. 5: Mechanistic experiments.

a–b The effect of K₂CO₃ and attempted cycloaddition between 1a and 14. c Three-component reaction of 1a, 3a and alkynyl copper intermediate 2a-Cu. d Analysis the reaction progress by ¹H NMR.

Reaction progress kinetic analysis (RPKA), following the approach introduced by Blackmond et al., was conducted using calorimetric measurements to continuously monitor reactions under synthetically relevant conditions^58,59,60. In this three-component reaction, a kinetic isotope effect (KIE) of 0.84 was observed upon using deuterated alkyne 2a-d in comparison with 2a, indicating an inverse secondary isotope effect (Fig. 6a, and Supplementary Fig. S2)⁶¹. This observation implies that the rate-determining step involves changes in the hybridization of the carbon atom attaching hydrogen/deuterium. In contrast, the [3 + 2] cycloaddition between 2a/2a-d and 3a exhibited a KIE of 1.44 (Supplementary Fig. S3), suggesting that C–H bond cleavage is likely rate-limiting step under those conditions. These results highlight that using the diaryliodonium reagent to quench the copper triazolide intermediate fundamentally alters the rate-determining step relative to a pathway where the copper species is terminated through protonation (Supplementary Figs. S4–S8). Further calorimetric studies evaluating each substrate’s impact on the reaction revealed first-order kinetics in both 2a and 3a (Fig. 6b). However, a formal 0.4th-order dependence on 1a was observed, suggesting that the reaction rate is possibly governed by multiple transient intermediates in this multi-step process. Notably, product inhibition prevented complete conversion at low copper catalyst loadings (see below), which complicated detailed kinetic analysis. Nevertheless, similar analysis of the copper catalyst gave a rough conclusion: copper catalyst displayed concentration-dependent behavior. First-order dependence on [Cu] was observed at loadings below 1.0 mol%, but shifted to zeroth-order kinetics when the catalyst loading exceeded 1 mol%.

Fig. 6: Kinetic studies.

a Kinetic isotope effect analysis with 2a and 2a-d. b Measuring the orders of three different substrates.

During our investigation, we found that reducing the copper catalyst loading (e.g., to 2–5 mol%) led to incomplete conversion and a slight decrease in enantioselectivity. We hypothesized that the product or byproducts of this reaction might cause catalyst deactivation or inhibition, or promote the formation of off-cycle copper species, thus reducing the concentration of active copper. To explore this possibility, we added one extra equivalent of product 4a to the interrupted CuAAC reaction and observed a noticeably slower reaction rate (red curve vs. blue curve in Fig. 7 and Supplementary Fig. S9.). These results suggest that the triazole moiety of 4a may coordinate to the copper atom, thereby diminishing the [Cu]_active.

Fig. 7

Graphical rate equations of the standard three-component reaction and the reaction with additional product 4a.

Based on the above experimental observations and relevant literatures, we propose a possible mechanism for the cyclic diaryliodonium-interrupted CuAAC reaction (Fig. 8). In the two-component CuAAC reaction between benzoylacetylene 2a and azide 3a, it is generally accepted that the rate-determining step is the protonation step of triazolyl copper with the alkyne, in line with existing CuAAC kinetic studies (Fig. 8a). In 2013, Fokin et al. extensively investigated the formation of a copper triazolide intermediate from an alkyne and an azide, identifying a key dinuclear copper species^62,63. However, in our catalytic system, we suggested that an alkynyl copper intermediate RC≡CCu(I) may be not the key intermediate, which is not always necessary for CuAAC reaction (Fig. 8b)⁶⁴. Instead, Cu(I)L* may react both benzoylacetylene and azide in a concerted cyclization to give intermediate M3, involving conversion of a C(sp)–H into a C(sp²)–H bond. Under basic conditions, M3 undergoes an elimination followed by rearrangement, potentially via a vinylidenyl copper species, to generate M2. Next, the cyclic diaryliodonium undergoes oxidative addition with M2, furnishing a trivalent copper species M4 that incorporates an axially chiral framework; this step determines the reaction’s stereoselectivity. Finally, reductive elimination from M4 delivers the product 4a and regenerates the Cu(I) catalyst. From our kinetic studies, several points merit further explanation: (a) The rate of copper triazolide quenching by the cyclic diaryliodonium exceeds the protonation rate (k₄ > k₃). (b) The rates k₁ and k₄ are likely comparable and collectively influence the overall reaction rate, as evidenced by two observations: (i) an inverse secondary isotope effect for the terminal alkyne C-H bond, and (ii) an apparent 0.4th-order dependence on the cyclic iodonium. (c) The product may coordinate to Cu(I)L* in a coordination–dissociation equilibrium, thereby causing product inhibition in the reaction.

Fig. 8: Plausible catalytic cycle.

a Pathway for Classic CuAAC in Aprotic Solvent. b Rationale for Cu-catalyzed Coupling of Alkyne, Azide and Diaryliodonium.

In summary, we have developed an asymmetric interrupted CuAAC reaction by capturing the copper triazolide intermediate with cyclic diaryliodoniums, thereby providing an efficient route to a variety of atropisomeric biaryl triazoles and significantly expanding the CuAAC repertoire. Moreover, this work also marks an application of formal carbon nucleophiles in a copper-catalyzed atroposelective ring-opening of cyclic diaryliodoniums for atropisomer synthesis. Mechanistic studies uncovered three key insights: (a) quenching of the copper triazolide intermediate by the cyclic diaryliodoniums outcompetes the protonation reaction by terminal alkynes⁶⁵; (b) an inverse secondary kinetic isotope effect for the terminal alkyne indicates that the conversion of its terminal carbon center from sp to sp² hybridization is the rate-determining step; and (c) adding extra triazole product to the standard reaction mixture notably diminishes the reaction rate, implying that triazole effectively competes with pyridine-2,6-bis(oxazoline), e.g., L1, for copper coordination. Consequently, designing ligands that coordinate more strongly to copper—and thus mitigate triazole’s competitive binding—may offer alternative pathways to lower catalyst loadings and enhance stereoselectivity in our future research.

Method

General procedure for Cu-catalyzed coupling of cyclic iodonium, propynones, azidoacetates in CH₂Cl₂

4 Å Molecular sieves (80 mg) in a Schlenk tube were dried by a heating gun under vacuum at 400 °C for 5 min. After cooling down, Cu(MeCN)₄PF₆ (3.7 mg, 0.01 mmol, 10 mol%), L1 (7.4 mg, 0.02 mmol, 20 mol%) and anhydrous CH₂Cl₂ (2.0 mL) were added, the resulting mixture was stirred for 30 min. Cyclic diaryliodonium salt 1 (0.10 mmol, 1.0 equiv.), propynone 2 (0.12 mmol, 1.2 equiv.) and azidoacetate 3 (0.12 mmol, 1.2 equiv.) were added. After stirring at 25 °C for 30 seconds, potassium carbonate (41.5 mg, 0.30 mmol, 3.0 equiv.) (pre-dried by a heating gun under vacuum at 400 °C for 3 minutes) and anhydrous CH₂Cl₂ (2.0 mL) were added and stirred at 25 °C for 36 h. After complete consumption of diaryliodonium, the mixture was filtered through a plug of celite with ethyl acetate. The filtrate was concentrated under reduced pressure, and the residue was purified by column chromatography on silica gel to deliver the corresponding product.