Catalytic asymmetric functionalization of bicyclo[1.1.0]butane boronic esters enabled by 1,2-oxygen migration

Abstract

The formation of boronate complexes through the coordination of alkoxide salts with organoborons serves as key intermediates in transition-metal-catalyzed coupling reactions. These complexes typically undergo classical transmetallation followed by reductive elimination to form C–C and C–X bonds. However, 1,2-oxygen migration in such boronate complexes remain underdeveloped, particularly in asymmetric catalysis, likely due to competing transmetallation pathways. Successfully achieving this migration potentially establish a fundamentally distinct reaction step, thereby significantly broadening the scope of transition-metal-catalyzed coupling reactions. In this study, we develope an iridium-catalyzed asymmetric alkylation that enables the stereoselective construction of cis-cyclobutanes. Our strategy employs the high ring strain of bicyclo[1.1.0]butane (BCB)-derived boronate complexes to induce 1,2-alkoxy migration. Mechanistic investigations reveal that 1,2-alkoxy migration proceeds stepwise, beginning with C–C bond formation, followed by a distinctive C–B bond rotation and subsequent 1,2-oxygen migration. Notably, exocyclic C–B rotation, which is outcompeted by 1,2-carbon migration in prior studies, can occur in this system because the competing 1,2-oxygen migration has a higher energy barrier. These findings provide crucial insights into the origin of stereoselectivity and the rate-determining step of this reaction.

The formation of alkoxyl boronate complexes in cross-coupling reactions typically undergo classical transmetallation followed by reductive elimination to form C–C and C–X bonds. Here, the authors report an iridium-catalyzed asymmetric alkylation that enables the stereoselective construction of cis-cyclobutanes, employing the high ring strain of bicyclo[1.1.0]butane-derived boronate complexes to induce 1,2-alkoxy migration.

Introduction

Four-membered cyclobutanes are important structural motifs found in numerous small-molecules and pharmaceuticals, exhibiting a broad range of biological and medicinal activities^1,2. Over the past years, various innovative methodologies have been developed for their synthesis, significantly advancing this field and enabling the efficient construction of complex molecular architectures^3,4. Among these structures, functionalized cyclobutanols represent significant scaffolds in bioactive molecules (Fig. 1a)^5,6. Traditionally, their synthesis has relied heavily on the addition of organometallic reagents to cyclobutanones. Recently, the development of catalytic methods, particularly those involving transition metals^7–9, and radical-mediated pathways¹⁰, has further advanced this area. However, catalytic asymmetric synthesis of cyclobutanol-containing chiral compounds remains underdeveloped^11–14.

Fig. 1. The significance of cyclobutanols and the conceptual design of this work.

a Cyclobutanols in bioactive molecules. b Transition-metal-catalyzed Suzuki-Miyaura cross-coupling reaction. c Transition-metal-catalyzed asymmetric cross-coupling via base-activated organoborons. d Asymmetric 1,2-oxygen migration (stepwise mechanism).

Organoborons are indispensable building blocks in organic synthesis, medicinal chemistry, and material science^15–19. They are widely utilized in classical transition-metal-catalyzed cross-coupling reactions^20–22. In such transformations, the empty p-orbital of the boron center readily coordinates with basic alkoxides to generate tetra-substituted boronate complexes, which then undergo facile transmetallation followed by reductive elimination to furnish new C–C or C–X bonds. Notably, alkoxyboronate byproducts are formed during this process, as exemplified by the classic Suzuki–Miyaura coupling reaction (Fig. 1b). In contrast, harnessing these boronate complexes to promote 1,2-oxygen migration offers a unique opportunity to deepen our understanding of fundamentally distinct reactions and potentially enable the development of entirely different coupling transformations. Since Morken’s pioneering studies on asymmetric conjunctive coupling reactions^23–27, later independently developed by Ready^28–31, Aggarwal^32,33, Song³⁴, our group^35–38, and others^39–42, significant progress has been made in stereoselective 1,2-metallate rearrangements, providing access to chiral organoborons. Nevertheless, most reported examples have focused on 1,2-carbon migration^43–47, whereas advances in asymmetric heteroatom migration remain limited³⁵. To the best of our knowledge, Trost and Zhang independently developed the 1,4-oxygen shift enabled by π-allylic palladium complexes^48,49. However, catalytic asymmetric 1,2-oxygen migration remains undeveloped, likely due to direct competition with the well-established transmetallation pathway (Fig. 1c).

Bicyclo[1,1.0]butanes (BCBs) have attracted considerable attention due to their unique reactivity^50–55, particularly when bearing electron-withdrawing groups that direct nucleophilic or radical additions to the β-position^56–65. Moreover, BCBs have demonstrated potential in strain-release-driven 1,2-metallate shift reactions from BCB-derived boronate complexes^66–69. However, catalytic asymmetric variants of these reactions remain elusive. Inspired by Ready’s studies on Ir-catalyzed asymmetric migratory reaction^29–31, a seminal report described an iridium-catalyzed asymmetric allylation mediated by ring-strained B-ate complexes, where the reaction proceeds through either a stepwise or concerted pathway, preferentially affording the trans-diastereomers through 1,2-carbon migration³². Building on our interest in asymmetric migratory coupling reactions^35–38, we envisioned that alkoxylithium reagents could be employed to activate BCB-derived boronic ester in situ. However, it remained unclear whether the inherent ring strain of BCBs could suppress the competing transmetallation pathway and selectively promote the underexplored asymmetric 1,2-oxygen migration. Furthermore, since alkoxide salts are prepared from alcohols in situ, the residual protons might facilitate undesired hydroalkoxylation of the C–C σ bond in BCBs⁶⁸. Additionally, chiral ligands are required to modulate the isomerization rate of Ir^III(π-allyl) intermediate^70–75, thereby enabling its interception by the migratory species of the ring-strained B-ate complex. If these challenges can be addressed, this strategy could ultimately enable the formation of chiral cyclobutanol derivatives via carboalkoxylation of the C–C σ bonds in BCBs (Fig. 1d).

Results and discussion

Optimization of the reaction conditions

To test this hypothesis, we initially investigated the reaction using a catalyst system derived from [Ir(COD)Cl]₂ and the Carreira ligand (R_a)-L1^74,76–83. The oxygen-coordinated boronate complex was prepared in situ by treating BCB-derived boronate with lithium alkoxide 1a. This reaction proceeded successfully, affording the desired migratory coupling product 4a in 32% yield, 97% ee, and 17:1 diastereomeric ratio (Table 1, entry 1). It is noteworthy that the syn-diastereomer was observed, in contrast to the previous 1,2-carbon addition, which resulted in anti-diastereomer³². Further optimization of the stoichiometric ratio between lithium alkoxide 1a and boronic ester 2 revealed that a 1.5:1 ratio (1a:2a) significantly improved the yield and minimized byproduct formation compared to other ratios tested (entries 2–6). The addition of excess lithium alkoxide proved essential for the efficient formation of the boronate complexes. Furthermore, a series of phosphoramidite ligands, either through functional group modifications of the Carreira ligand or altering the chiral scaffold, were systematically evaluated (entries 7–10). However, these modified ligands exhibited lower catalytic efficiency and failed to effectively promote the desired transformation. Consequently, the optimal reaction conditions were established as shown in entry 5: employing 1.5 equivalents of 1a and 1.0 equivalent of 2a in the presence of [Ir(COD)Cl]₂ (2.5 mol %) and (R_a)-L1 (10 mol %) in a 2:1 mixture of DCM and MeCN at 40 °C under nitrogen atmosphere.

Table 1.

Investigations of reaction conditions^a

Entry	Ligand	1a:2a:3a	Yield (%)b	ee (%)c	drd
1	L1	1:1:2.1	32	97	17:1
2	L1	1.2:1:2.1	35	97	17:1
3	L1	1.3:1:2.1	78	97	17:1
4	L1	1.4:1:2.1	85	98	17:1
5	L1	1.5:1:2.1	96 (86)	98	16:1
6	L1	1.7:1:2.1	93	97	16:1
7	L2	1.5:1:2.1	75	94	17:1
8	L3	1.5:1:2.1	86	97	16:1
9	L4	1.5:1:2.1	28	96	16:1
10	L5	1.5:1:2.1	18	95	16:1

^a1a, 2a, 3 (0.42 mmol, 2.1 equiv.), [Ir(COD)Cl]₂ (2.5 mol %) and ligand (10 mol %) in DCM (dichloromethane) and MeCN (acetonitrile) at 40 °C.

^bDetermined by ¹H NMR analysis using CH₂Br₂ as an internal standard, isolated yield shown in parenthesis.

^cDetermined by HPLC analysis.

^dDetermined by ¹H NMR analysis.

Substrate scopes

Under the optimized conditions, the substrate scope of complex lithium alkoxides was initially explored (Fig. 2). The prevalence of oxygen-containing functional groups in biologically active compounds such as sugars, ribose, natural products, and pharmaceuticals underscores their significance. This method was successfully applied to complex lithium alkoxides, enabling the straightforward synthesis of bioactive, oxygen-functionalized molecules featuring a cyclobutane moiety. To our delight, diastereoselective 1,2-oxygen migration was successfully achieved for pyranose, furanose and sugar derivatives, affording homologated products containing cis-cyclobutane units (4b–j, 47–68% yields, 4:1–17:1 dr). Notably, the complex structure and stereochemical features of sugars did not significantly interfere with the reaction’s activity and selectivity. Lithium alkoxides derived from valinol and naproxen also served as viable precursors for 1,2-oxygen shift (4k–l). The broad applicability of this approach was further demonstrated by its success with substrates derived from terpenes and steroids, furnishing products 4m–p in good yields with high diastereoselectivity. Of particular interest, a substrate containing a tertiary C–O bond afforded a nearly single diastereomer in 70% yield, with the reaction proceeding efficiently despite the steric hindrance of the migrating group (4n). Notably, the reported dr values may correspond to the ratios of diastereomers defined by the newly formed stereogenic elements (i.e., the cis/trans relationship on the cyclobutane ring and the stereogenic center), in combination with the pre-existing stereocenters of the natural products. Because the original stereocenters of the natural products remain intact, the reaction could, in principle, afford four possible diastereomers. However, only two diastereomers are observed, owing to the high level of stereocontrol imparted by the chiral catalyst, as the allylic iridium intermediate is typically formed in near-enantiopure form. Consequently, the observed diastereomeric ratios primarily reflect the cis/trans selectivity of cyclobutane ring formation. This highly enantio- and diastereoselective migratory process proved effective across a wide range of structurally diverse alkoxides. Aliphatic acyclic alcohol derivatives, including ethyl, iso-propyl and tert-butyl, afforded the desired products in 69–85% yields (4q–s) with high enantioselectivity. Higher diastereoselectivity was observed with increasing steric hindrance of the substituents on the migratory groups. Moreover, allylation products with varying ring sizes were synthesized, including six-membered (4t), seven-membered (4u), and even twelve-membered (4v) rings. A sterically bulky tertiary adamantanol was also successfully employed in the 1,2-oxygen shift, affording compound 4w in 52% yield with >20:1 dr.

Fig. 2. Substrate scope of lithium alkoxides^a,b'.

^a1 (0.4 mmol, 2.0 equiv.), 2a (0.2 mmol, 1.0 equiv.), 3a (0.42 mmol, 2.1 equiv.), [Ir(COD)Cl]₂ (2.5 mol %), (R_a)-L1 (10 mol %) in DCM (dichloromethane) and MeCN (acetonitrile) at 40 °C. ^bUsing 1.2 equivalents of 1r. ^b'Isolated yield.

Next, we examined a series of branched allylic carbonates 3 using the lithium alkoxide-activated BCB-derived boronate complex I, prepared in situ (Fig. 3). Ortho-, meta-, and para-substituted aromatic rings in 3, bearing electron-donating methyl (4x–za) and methoxy (4zb–zd) groups, were well-tolerated, affording the corresponding products in 70–85% yields and 92–98% ee with high diastereoselectivity. Allylic carbonates containing halogen substituents, including fluoro (4ze and 4zf), chloro (4zg), bromo (4zh and 4zi), and iodo (4zj) groups, also reacted smoothly, delivering the desired migratory products in good yields and high stereoselectivity. Notably, retention of bromine and iodine, which typically interfere with oxidation addition in transition-metal catalysis, enables the precise construction of molecular diversity through late-stage transformations. This reaction was also compatible with strong electron-withdrawing groups, such as trifluoromethyl (4zk), pinacol boronic ester (4zl), and cyano (4zm and 4zn) groups. Despite the propensity of such substrates to form unstable allylic cationic intermediates, the corresponding products were obtained in synthetically useful yields. Furthermore, substrates bearing electron-neutral substituents, including 4-biphenyl, 1-naphthyl and 2-naphthyl groups, afforded cis-cyclobutane chiral boronates in 72–76% yields and 94–98% ee. A diverse range of heteroaromatic rings, including furan (4zr), thiophene (4zs–zt), benzofuran (4zu), benzothiophene (4zv), and pyridine (4zw), were successfully incorporated into 1,2-oxygen migratory reactions. Notably, this strategy also proved effective for alkenyl- and alkyl-substituted carbonates, even in the presence of potentially unstable allylic iridium intermediates (4zx–zy). The allylic carbonate with 1,3-diphenyl substituents afforded the desired product 4zz in 42% yield with 81% ee and 15:1 dr. However, this reaction was incompatible with other allylic carbonates (for details, see Supplementary Section 5 in the Supplementary Information). Furthermore, we investigated the methyl-substituted BCB-derived boronic ester 2b. Due to the increased steric hindrance, a longer reaction time was required to improve conversion, with minimal impact on stereoselectivity (4zza–zzb). Finally, the reaction of the phenyl-substituted BCB derivative 2c afforded the corresponding product 4zzc in nearly enantiopure form.

Fig. 3. Substrate scope of allylic carbonates^a,b.

^a1a (0.3 mmol, 1.5 equiv.), 2 (0.2 mmol, 1.0 equiv.), 3 (0.42 mmol, 2.1 equiv.), [Ir(COD)Cl]₂ (2.5 mol %), (R_a)-L1 (10 mol %) in DCM (dichloromethane) and MeCN (acetonitrile) at 40 °C. ^bIsolated yield.

Mechanistic studies

To elucidate the mechanism of iridium-catalyzed allylation mediated by 1,2-oxygen shift, we conducted a series of mechanistic studies (Fig. 4). We first investigated whether kinetic resolution of racemic carbonate 3a occurs in the reaction with BCB-derived boronate complex I. This approach enabled the isolation of the oxygen-transferred allylated cyclobutane product 4a, along with the recovery of unreacted 3a (Fig. 4a). The recovered carbonate, (S)−3a, was isolated in 50% yield with 90% ee, indicating a kinetic resolution process. To further support the proposed mechanism, we monitored the reaction of enantiomerically enriched allyl carbonate (S)-3a using ligands (S_a)-L1, (R_a)-L1, and rac-L1 independently (Fig. 4b). Employing ligand (S_a)-L1 afforded the product in 80% yield with good enantioselectivity, along with 39% recovery of (S)-3a. In contrast, use of (R_a)-L1 led to only 22% yield of the oxygen-transfer coupling product, accompanied by a significant decrease in enantioselectivity. The enantiomeric excess of the recovered carbonate also slightly decreased to 75%. Using the racemic ligand rac-L1 resulted in a product with slightly lower enantioselectivity compared to (S_a)-L1. These results suggest that the reaction of racemic allylic carbonate with the Ir(P, olefin) complex proceeds via kinetic resolution, generating matched and mismatched diastereomeric Ir^III(π-allyl) intermediates. A positive non-linear effect observed between the enantiomeric excess of the product and the enantiopurity of the ligand suggests that the stereo-determining steps involve two chiral ligands and a homochiral metal–ligand complex might be more reactive than the heterochiral combination (Fig. 4c)⁸⁴. To investigate the influence of substituents on the allylic carbonates in the oxygen-transfer coupling reaction, we tested carbonates bearing both electron-donating and electron-withdrawing groups (Fig. 4d). The resulting Hammett plot revealed a linear correlation between the relative reaction rate and the corresponding σ+ value (ρ = −0.77), implying the formation of a positive charge on the electrophilic carbon in the rate-limiting step. This suggests that electron-rich substituents stabilize the positive charge in the Ir^III(π-allyl) species. Furthermore, density functional theory (DFT) calculations at the M06-2X/def2-TZVP (SMD = MeCN)//M06-2X/def2-SVP level were performed to investigate the reaction mechanism and the stereoselectivity of the strain-release-driven C–C and C–O bond formation (Fig. 4e). Our calculations begin with the formation of a cationic Ir^III(π-allyl) complex (int1) via an S_N2’-type oxidative addition of the allylic carbonate, a well-documented process in the literature³⁰. Subsequently, int1 reacts with the BCB-Bpin (int2) via transition state TS1, which has an activation barrier of 13.0 kcal/mol, yielding int3. The formation of int4 proceeds via TS2, which exhibits a higher energy barrier (ΔΔG^‡ = 3.6 kcal/mol) compared to that of forming int3, indicating a less favorable pathway. In the more favorable C–C bond formation transition state TS1, the OⁱPr group occupies a less sterically hindered quadrant, minimizing steric interactions with the Carreira ligand. Thus, the formation of int4 via TS2 is both kinetically and thermodynamically disfavored. Surprisingly, rotation of C–B bond in int3 via TS3, which has been considered challenging in prior literature³², is found to be facile, with a low barrier of only 4.3 kcal/mol, leading to int4. The subsequent 1,2-oxygen migration can proceed via either TS4 or TS5, with TS4 being more favorable (−7.3 kcal/mol vs. −5.7 kcal/mol). Notably, the 1,2-oxygen migration via TS5 directly from int3 is 1.6 kcal/mol higher in energy than the pathway from int4, which is formed through a distinctive C–B bond rotation of int3. The significant steric clash between the isopropoxy group and the phenyl group on the ligand renders TS5 energetically unfavorable. To elucidate the impact of this steric effect, transition states for the 1,2-methoxy migration were calculated for comparison. These calculations revealed identical activation barriers for the 1,2-oxygen shift process (see Fig. S14 in the Supplementary Information for details), consistent with the observed decrease in diastereoselectivity for the substrate bearing an ethoxy group with reduced steric hindrance. It is worth noting that the formation of int5 is more favorable than that of int6, ultimately leading to the desired cis-product through dissociation of the iridium complex.

Fig. 4. Mechanistic studies.

a Kinetic resolution of racemic carbonate. b Matched and mismatched reactions with (S)−3a. c Nonlinear effect. d Hammett plot. e DFT calculations. f Kinetic studies. g Intermolecular competition experiments. DCM: dichloromethane. MeCN: acetonitrile. THF: tetrahydrofuran.

Further mechanistic insights were obtained through kinetic studies to assess the influence of each component, including the boronate complex, the Ir/L catalyst, and the allylic carbonate, on the reaction rate. The results revealed first-order kinetics with respect to both the boronate complex and the catalyst (Fig. 4f). In contrast, the zero-order dependence on the allylic carbonate suggests that the oxidative addition between the Ir/L catalyst and the allylic carbonate is not the rate-limiting step. Instead, the Ir^III(π-allyl) intermediate is involved in the rate-determining step, as the reaction rate varies with the electronic nature of the carbonate (Fig. 4d). Based on these kinetic studies in combination with DFT calculations, we propose that the reaction of int1 with int2 via TS1 constitutes the rate-determining step, involving both the allyliridium complex and the boronate complex. To further probe this, we employed two different alkoxylithium reagents with contrasting electronic properties to prepare tetra-substituted boronate complexes. When reacted with the standard substrate 3a, the complex bearing a terminal methyl group reacted approximately twice as fast as the one bearing a trifluoromethyl group (4zzd vs. 4zze). These results further support that C–C bond formation via TS1 is the rate-determining step, since the stabilized allylic carbocation intermediate undergoes a faster reaction with the more electron-rich B-ate complex (Fig. 4g).

Late-stage transformations of the desired product

To demonstrate the compatibility and synthetic utility of this methodology, we scaled up the reaction and carried out diverse late-stage molecular transformations, as shown in Fig. 5. To our delight, this reaction proceeded smoothly on a 2 mmol scale, delivering the target product 4zo in 86% isolated yield without any erosion of stereoselectivity. The absolute configuration of 4zo was determined to be R with cis-configuration on the cyclobutane ring, as confirmed by X-ray crystallographic analysis. The absolute configuration of the other products was assigned by analogy to 4zo. The product’s multiple derivatization sites enabled a broad range of subsequent transformations. Pd/C hydrogenation and Ir-catalyzed hydroboration of olefins furnished products 5 and 6 in high yields while preserving stereoselectivity. Wacker oxidation and olefin metathesis

Fig. 5. Late-stage transformations.

a Scale-up reaction. b Diversifications of 4zo. DCM dichloromethane, MeCN acetonitrile, NBS N-Bromosuccinimide, LDA lithium diisopropylamide, THF tetrahydrofuran.

successfully afforded the desired products 7 and 8, respectively, and the pinacol boronic ester did not significantly affect these transformations. Reaction of 4zo with KHF₂ converted the Bpin group into the corresponding trifluoroborate potassium salt 9. A pyridine unit was stereoselectively introduced via a transition-metal-free cross-coupling reaction. Additionally, the tertiary boronic ester underwent Matteson homologation to afford a primary boronic ester, which was further oxidized to yield the primary alcohol 11. Furan and olefin moieties were introduced in high yields via Aggarwal’s arylation using 2-furanlithium reagent and Zweifel olefination employing vinylmagnesium bromide, respectively^85,86. Remarkably, Zweifel olefination of 4zo followed by 1,2-elimination afforded alkyne 14 in 65% yield. Furthermore, treating 4zo with H₂O₂ and NaOH led to the selective formation of lactone 15 via oxidative rearrangement, while retaining the olefin moiety, albeit with 1:1 diastereoselectivity. Phenyllithium was prepared to activate 4zo via the formation of a boron-ate complex, followed by intramolecular nucleophilic substitution to afford the intermediate. This intermediate was used directly, without isolation, in the Zweifel reaction, affording 16 in 34% yield over two steps with 96% ee and 9:1 dr. Finally, a Grubbs olefin metathesis reaction delivered bicyclo[3.1.1]hept-2-ene 17 in 80% yield with 98% ee as a single diastereomer, likely because cis-16 is unable to undergo the metathesis reaction. These transformations further highlight the unique advantages of these products in constructing molecular complexity.

In summary, we have developed a highly efficient method for the enantio- and diastereoselective synthesis of chiral cis-cyclobutanol derivatives featuring C–B bonds. This methodology harnesses the strain-release of bicyclo[1.1.0]butane-derived B-ate complexes and proceeds via 1,2-oxygen migration mediated by asymmetric allylic substitution, avoiding the formation of classical transmetallation byproducts. The reaction follows a stepwise pathway, involving C–C bond formation, followed by a distinctive C–B bond rotation and subsequent 1,2-oxygen migration. This mechanism is distinct from previous reactions, which proceed through either a stepwise or concerted pathway, preferentially affording the trans-diastereomer. Notably, the C–B bond rotation in our reaction is a key factor that enables access to the cis-diastereomer.

Methods

General procedure for the preparation of cyclobutane boronic esters enabled by 1,2-oxygen migration

To a 10 mL flame-dried Schlenk-type sealed tube equipped with a Teflon-coated magnetic stir bar were added the corresponding alcohol 1 (0.40 mmol, 2.0 equiv.) and hexane (1.5 mL) under N₂ atmosphere. The solution was stirred thoroughly and cooled to −78 °C, followed by injecting ⁿBuLi (0.4 mmol, 2.0 equiv., 1.6 M in hexanes) slowly. The mixture was stirred at −78 °C for 15 min, then warmed to room temperature and stirred for an additional 30 min. The solvent was carefully removed under vacuum, and the residue was redissolved in anhydrous ether (1.0 mL). Bicyclo[1.1.0]butyl boronic ester 2a (36.0 mg, 0.2 mmol, 1.0 equiv.) was then added, and the solution was stirred at room temperature for another 30 min. The mixture was carefully concentrated under reduced pressure to produce a white borate salt. Meanwhile, the other Schlenk tube was filled with [Ir(COD)Cl]₂ (3.4 mg, 0.005 mmol, 2.5 mol %), (R_a)-L1 (10.2 mg, 0.02 mmol, 10 mol %) and DCM (0.7 mL). After stirring at room temperature for 20 min, the catalyst solution was transferred to the reaction tube, and the Schlenk tube was rinsed with MeCN (0.4 mL). The remaining MeCN (1.0 mL) and the corresponding allyl carbonate (0.42 mmol, 2.1 equiv.) were then added to the reaction tube. The reaction mixture was heated at 40 °C for 24 h, with progress monitored by TLC. The reaction was quenched by the addition of saturated aqueous NH₄Cl, and the mixture was extracted with EtOAc (3 × 10 mL). The combined organic layers were washed with brine, dried over anhydrous Na₂SO₄ and filtered. After removing the solvent under reduced pressure, the residue was purified by flash column chromatography on silica gel to afford analytically pure products 4.

Author contributions

D.-W. G. conceived and supervised the project. X.-Y. Z. and T.-G. D. performed the experiments and examined the substrate scopes. X.-Y. Z. and C.-L. J. conducted the mechanistic studies. D.-W. G. wrote the paper with the feedback from all authors.

Peer review

Peer review information

Nature Communications thanks Jiannan Zhao and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The authors declare that all the data that support the findings of this study are available within the article and Supplementary Information files. Crystallographic data for the structure reported in this article have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number 2446112 (4zo). These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. Source data are provided with this paper. All data are available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-69860-9.