Introduction

Cyclobutyl moieties are prominent in natural products and biologically relevant molecules (Fig. 1a)1,2. They frequently occur in the core structures of polycyclic rings and feature multiple stereogenic centers. This characteristic has led to the development of numerous elegant catalytic asymmetric methodologies towards the synthesis of chiral molecules containing cyclobutanes. These methods include [2 + 2] cycloaddition reactions3,4,5,6,7, C−H bond functionalization8,9, and selective transformations of cyclobutenes, and others10,11,12,13,14. Recently, bicyclo[1.1.0]butanes (BCBs) and its derivatives, known for their weak σ-bond due to high ring-strain energy, have become popular in synthetic chemistry15,16,17,18,19,20,21,22,23,24,25. These compound could be utilized to synthesize four-membered ring units, mainly through two approaches26,27. Firstly, BCBs with an electron-withdrawing group at the terminal position of the C–C σ bond engage in nucleophilic or radical addition at the β-position of the bridging bond, leading to the formation of multi-substituted cyclobutanes28,29,30,31. Additionally, the Aggarwal group recently developed strain-release reactions from BCB-derived B-ate complexes through either radical or ionic pathways, enabled by a 1,2-metalate shift (Fig. 1b)32,33,34. Although BCBs are versatile in chemical transformations for constructing cyclobutanes, asymmetric regulation of these reactions has progressed slowly, likely due to the challenge of suppressing background reactions caused by their high reactivity.

Fig. 1: Overview of previous studies and this work.
figure 1

a Chiral cyclobutanes in natural products and bioactive drugs. b Selective examples to construct cyclobutanes from BCBs. c Asymmetric conjunctive couplings via 1,2-metallate shift. d Asymmetric synthesis of biaryl atropisomers containing cyclobutylboronic esters (this work). ent-4a: enantiomer of 4a.

Full size image

In recent years, catalytic asymmetric 1,2-metalate shift has attracted extensive research interests due to their considerable synthetic potential. The enantioselective conjunctive coupling reactions, developed by the Morken group through 1,2-carbon migration, represent a notable advancement35,36. Subsequently, several groups have independently made significant contributions to the synthesis of chiral boronates through palladium37,38,39,40, nickel catalysis41,42,43 and organocatalysis44,45,46,47. Since 2021, the Ready group48,49,50, later by our group51,52, have developed Ir-catalyzed asymmetric allylic alkylation of diverse B-ate complexes through 1,2-carbon or boron shift migrations53,54,55,56,57,58,59. These methods delivered an array of chiral boronic esters with 1,3-stereocenters or gem-diborylalkenes. Despite these advancements, the application of asymmetric 1,2-metallate shift to ring-strained B-ate complexes remained unreported until the seminal research by the Aggarwal group. They disclosed an iridium-catalyzed asymmetric allylation enabled by strain release in BCBs, producing cyclobutanes with center chirality through the difunctionalization of C−C σ-bonds in B-ate complexes60. However, most of the reactions mentioned above are concentrated on producing central chirality through a 1,2-metallate shift. Considering the significance of biaryl atropisomers61,62,63 and cyclobutanes in both pharmaceutical and synthetic chemistry, the use of ring-strain boronate complexes as nucleophiles for constructing atropisomers is highly desirable (Fig. 1c).

Recently, dynamic kinetic asymmetric transformations (DYKATs) have attracted significant attention due to their wide applications in the synthesis of chiral axial biaryls64,65,66,67,68,69,70,71,72,73. Insipired by these seminal research, we envisioned that the diastereomeric Pd(II) intermediates would need to undergo rapid isomerization, followed by the capture of species generated from 1,2-carbon migration of ring-strained boronate complexes. This sequence enables an efficient dynamic kinetic asymmetric transformation to proceed smoothly, which is cruial for achieving high enantioselectivity. Furthermore, the bulky palladium(II) complex cannot approach from the endo side because of steric hindrance caused by the pinacol group and electronic effects, with higher electron density on the exo face of the β-carbon according to the previous studies32,74. Notably, the choice of an appropriate chiral catalytic system can influence the rate of migratory group transfer and the isomerization of biaryl Pd(II) intermediates, thereby affecting the stereoselectivity of this reaction (Fig. 1d). In this work, we have successfully applied this approach by using ring-strained boronate complexes as nucleophiles to construct atropisomeric cyclobutanes, which has never been realized. The C−B bond in the target would significantly facilitate subsequent molecular elaborations, allowing the construction of a library of axially chiral cyclobutanes.

Results and discussion

Optimization of the reaction conditions

To evaluate the feasibility of this reaction, we prepared a ring-strained BCB boronate complex, derived from PhBpin (2a) and bicyclo[1.1.0]butyl sulfoxide (1) in Fig. 2. We were pleased to find that the desired product 4a was obtained in 40% yield and 29% ee through Pd-catalyzed migratory coupling of naphthylisoquinoline trifluoromethanesulfonate (3a) with B-ate complexes. Only the cis-cyclobutane isomer was observed in this reaction. Further screening of another modified phosphine-oxazoline ligand L2, failed to yield a significant improvement. However, upon switching to a series of chiral bisphosphine ligands, we observed a notable enhancement in enantioselectivity with L7 proving to be the most effective, although the yield still has potential for improvement. Remarkably, no desired product was observed when (Ra)-DTBM-Segphos was used, which might be influenced by the sterically hindered tBu group on L5. On this basis, we have carefully screened the palladium source and the ratio of the components, finding that the yield can be further improved to 90%. Finally, when toluene is used as the solvent for catalyst coordination, we can obtain the product 4a in >95% NMR yield (82% isolated yield) with 93% ee.

Fig. 2: Investigations of reaction conditions.
figure 2

a1 (0.2 mmol, 1.0 equiv.), 2a (0.2 mmol, 1.0 equiv.), tBuLi (0.2 mmol, 1.0 equiv.), 3a (0.4 mmol, 2.0 equiv.), Pd2(dba)3 (0.005 mmol, 2.5 mol%) and ligand (0.012 mmol, 6 mol%) in THF at 60 oC. Only cis-cyclobutane isomers were observed in this reaction unless otherwise noted. bDetermined by 1H NMR analysis using CH2Br2 as an internal standard. cDetermined by HPLC analysis. dEnantioselectivity was not determined. ePd2(dba)3, 98%, ≥22.7% (as Pd). f1 (0.4 mmol, 2.0 equiv.), 2a (0.4 mmol, 2.0 equiv.), tBuLi (0.4 mmol, 2.0 equiv.), 3a (0.2 mmol, 1.0 equiv.), Pd2(dba)3 (0.005 mmol, 2.5 mol%) and (Ra)-L7 (0.012 mmol, 6 mol%) in THF at 60 oC. g1 (0.4 mmol, 2.0 equiv.), 2a (0.4 mmol, 2.0 equiv.), tBuLi (0.4 mmol, 2.0 equiv.), 3a (0.2 mmol, 1.0 equiv.), Pd2(dba)3 (0.005 mmol, 2.5 mol%) and (Ra)-L7 (0.012 mmol, 6 mol%) in toluene at 60 oC. The isolated yield is 82%. L3: (Ra)-BINAP; L4: (Ra)-Segphos; L5: (Ra)-DTBM-Segphos.

Full size image

Substrate scopes

As presented in Fig. 3, under the optimized reaction conditions, the variation of functional groups on the pyridine ring, including both electron-donating (3b–c: CH3) and electro-withdrawing groups (3d–e: F or CF3), had negligible effects on the reactivities and enantioselectivities. This led to the production of the corresponding axially chiral cyclobutyl boronic esters in good yields with excellent enantioselectivities (92–99% ee). The absolute configuration of 4c was determined through x-ray crystallographic analysis. Further screening of substituents with varying electronic properties on the naphthyl/phenyl group of the isoquinoline-derived OTf was also conducted. It was observed that the mild reaction conditions were compatible with a wide range of substituents, including methoxy (3f), methyl (3g), cyano (3h) and ester (3i) groups. The quinazoline and pyrimidine moieties were also well accommodated, leading to the desired products in 56% and 77% yields, respectively, with uniformly high enantioselectivities (4j–k). Moreover, it is noteworthy that axially chiral compounds 4l–m, each featuring two cyclobutane moieties, were synthesized in 48% yield with 87% ee and 49% yield with 82% ee, respectively, through Pd-catalyzed stereoselective migratory coupling, accompanied by further coupling of the bromide tethered to the naphthyl ring. Subsequently, we initiated investigations on organoboronic esters with a variety of aryl or alkenyl substituents. We tested substrates featuring a methyl group at various positions on the phenyl ring, all of which afforded satisfactory results (4n–p). A broad range of aryl boronic esters, including those with electron-donating (2q–s), electron-withdrawing (2t–w), and electron-neutral groups (2x–za) were effectively incorporated, resulting in the synthesis of axially chiral molecules with cyclobutyl boronic esters. These molecules were synthesized in yields ranging from 34% to 85% yields and exhibited enantiomeric excesses (ees) between 88% and 99%. Additionally, a diverse set of heterocycles proved to be compatible, such as thiophenes, benzofuran and benzothiophenes at different positions, as well as indole, all of which delivered synthetic useful yields and high stereoselectivity (4zb–zh). Despite the potential for competitive reactions involving 1,2-carbon migration to either the sigma bond of strained rings or olefin groups in these substrates (2zi–zl), the reaction was observed to selectively migrate to the C–C σ bond of strained B-ate complexes instead of the C=C bonds. It should be noted that only one cis-cyclobutane isomers were produced in the migratory coupling reactions of the aforementioned substrates. Subsequently, we attempted to use natural product-derived boronic esters, including those derived from Naproxen, terpenoids, and steroids. When using the boronic ester derived from Naproxen as the substrate, two diastereomers were isolated in a 3:1 ratio, with 85% isoalted yield. Terpenoid and steroid-derived borates were found to be well compatible with the reaction conditions, affording the target products in 32-81% isolated yields with diastereomeric ratio (dr) ranging from 5:1 to 15:1 (4zn–zq). We also investigated the effect of using a ligand with the opposite configuration by using 2zm as the substrate, which resulted in a reversal of diastereomer preference, with the minor diastereomer becoming the major one (4zm, 1:3 dr). Furthermore, we tested terpenoid- and steroid-derived boronic esters with ligands of opposite configurations, consistently observing a reversal in the diastereomeric ratio, where the minor diastereomers became the major ones. The yields of these reactions with (Sa)-L7 or L8 decreased slightly, likely due to mismatching with the chiral centers in the natural product-derived boronic esters. These findings suggest that ligand configuration significantly influences the diastereoselectivity of natural product-derived boronic acid esters, while substrate structure has minimal impact. We believe that the source of the diastereoselectivity in these products (4zm–zq) arises from the combination of axial chirality and the inherent chirality of natural product-derived boronates, rather than from cis- or trans-isomers of cyclobutanes. Notably, the cis isomer of cyclobutane was consistently observed since the bulky palladium(II) complex cannot approach from the endo side due to the steric hindrance and the electronic effects (For details, see section 8.9 of the Supplementary Information)32,74. These results further demonstrate the excellent compatibility of this method. It should be noted that the chiral centers in terpenoid- and steroid-derived borates are distant from the atropisomeric moiety, NMR analysis cannot discriminate the diastereomers. Therefore, chiral HPLC analysis was conducted to determine the diastereoselectivity (4zn–zq).

Fig. 3: Substrate scope of biaryl skeletons, aryl, vinyl and natural product derived boronic esters.
figure 3

a1 (0.40 mmol, 2.0 equiv.), 2 (0.40 mmol, 2.0 equiv.), 3 (0.20 mmol, 1.0 equiv.), Pd2(dba)3 (0.005 mmol, 2.5 mol%) and (Ra)-L7 (0.012 mmol, 6 mol%) in THF and toluene at 60 oC. Only cis-cyclobutane isomers were observed in this reaction unless otherwise noted. bUsing (Ra)-L6 in MeCN instead of toluene. cUsing 1 (0.60 mmol, 3.0 equiv.) and 2a (0.60 mmol, 3.0 equiv.). dReacting at toluene (0.6 mL) and MeCN (0.4 mL). eUsing (Ra)-L6 instead of (Ra)-L7. fUsing MeCN instead of toluene. gThe diastereoselectivity was determined by NMR analysis. hThe diastereoselectivity was determined by HPLC analysis. iUsing (Ra)-L8 in THF and DCM at 60 oC.

Full size image

We subsequently investigated alkyl boronic esters and found that both reactivity and enantioselectivity were lower compared to their aryl counterparts, possibly due to the differences in migration rates between alkyl and aryl groups. We then carefully examined the reaction conditions and discovered that using ligand L8, along with slightly modified reaction conditions (For details, see Supplementary Table 3), can achieve high levels of enantioselectivity with excellent yield in Fig. 4. The reaction of linear alkyl borates with biaryl OTf proceeded smoothly, producing the desired products with yields ranging from 78% to 88% and enantiomeric excesses of 91% to 93% (4zr–zt). This reaction was found to be compatible with bulky aliphatic boronic esters (4zu–zv). We also tested substrates containing benzyl, alkyl bromide, ether, alkenyl, ester and amide groups, which afforded axially chiral cyclobutane boronates with 40–72% yields and 89–98% ees (4zw–zzb). Simultaneously, we investigated a series of cyclic boronates, ranging from 3- to 6-membered rings, including a cyclohexyl borate containing a nitrogen unit, which delivered cis-isomers with synthetically useful yields and excellent chiral control (4zzc–zzf). Interestingly, 1,2-boron shift proceeds smothly to afford the atropisomeric gem-diborylcyclobutane in 82% yield and 88% ee with slightly modified reaction conditions using B2pin2 (bis(pinacolato)diboron) as the substrate.

Fig. 4: Substrate scope of alkyl boronic esters.
figure 4

a1 (0.40 mmol, 2.0 equiv.), 2 (0.40 mmol, 2.0 equiv.), 3a (0.20 mmol, 1.0 equiv.), Pd2(dba)3 (0.005 mmol, 2.5 mol%) and L8 (0.012 mmol, 6 mol%) in THF and DCM at 60 oC. Only cis-cyclobutane isomers were observed in this reaction unless otherwise noted. bUsing Pd(PPh3)4 and L12 in THF at 40 oC.

Full size image

Mechanistic studies

We studied the reaction mechanism. Firstly, chiral biaryl triflate 3a does not undergo racemization at 60 °C in THF and toluene, suggesting that the Pd(II) complex, formed through the oxidative addition of 3a, may undergo rapid isomerization, leading to a dynamic kinetic asymmetric transformation. In contrast, the hydrolysis byproduct 3a’ shows partial racemization at 60 oC. Since 3a’ is usually observed in the migratory coupling reactions, we propose that this byproduct is generated through the decomposition of 3a in the presence of B-ate complex II, which is in equilibrium with ArLi and its boron precursor. A 49% yield and 82% ee of 3a’ was obtained in the presence of II, likely due to the decomposition rate being slower than the racemization rate (Fig. 5a). Furthermore, no coupling product was observed from the racemic N-oxide 3zzh, indicating that the coordinating nitrogen atom of isoquinolyl/pyridyl unit may be required for the chelate-assisted oxidative addition step. Similarly, only trace amount of 4zzi was observed from (rac)-3zzi, which lacks a nitrogen atom, suggesting that the formation of a five-membered cationic palladacycle is essential for reactivity (Fig. 5b). When (Ra)-L7 was used as the chiral ligand, 43% yield and 91% ee were achieved, along with a significant decomposition of 3a likely due to the formation of a mismatched Pd(II) intermediate.

Fig. 5: Mechanistic studies.
figure 5

a Mechanistic investigations into the stablity of axial chiral molecules at 60 oC. b Studies on the effect of directing group. c Mechanistic studies on the effect of ligand. d Nonlinear effect.

Full size image

This intermediate can isomerize to the matched one, with the overall reaction rate being comparable to the hydrolysis rate. An 83% yield and −99% ee were obtained by employing (Sa)-L7 as the chiral ligand, indicating the formation of a matched Pd(II) intermediate from (Ra)-3a. (Ra)-3a’ was isolated in only 13% yield since the reaction rate of (Ra)-3a with palladium and (Sa)-L7 is much faster than its decomposition rate. When the racemic ligand was used, only 48% yield and −15% ee were observed. This suggests that the reaction rate of (Ra)-3a with (Sa)-L7 is only slightly faster than that with (Ra)-L7, resulting in a substantial amount of byproduct 3a’ (Fig. 5c). Finally, a study of the nonlinear effect was conducted, revealing a good linear correlation, which suggests that a monomeric palladium complex bearing a single ligand is involved in the enantioselectivity-determining step (Fig. 5d).

Late-stage transformations of the desired product

We scaled up the reaction to 2 mmol, and it is gratifying that this reaction generates the target product stereoselectively in 76% yield, preserving the enantioselectivity (92% ee) observed in the small-scale reaction. Based on this product, a variety of transformations were carried out as shown in Fig. 6. These include the simultaneous oxidation of the C–B bond and N atom of 4a to obtain the corresponding product 5. The Bpin moiety can be conveniently converted into the corresponding BF3 potassium salt quantitatively by reacting with KHF2. Additionally, 4a can undergo stereoselective arylation to obtain the corresponding products 7 and 8. One-carbon homologation can occur with CH2BrI, and the deboration of 4a proceeds smoothly to deliver the corresponding product 10. This reaction can also selectively undergo Zweifel olefination to afford the desired product 11 in 74% yield and 92% ee. Finally, vinyl carbamate was lithiated at the α-position with LDA and then reacted with 4a in Zweifel olefination, followed by treatment with tBuLi, which triggered the elimination of the resulting vinyl carbamate intermediate to form the alkyne product 12 in 54% yield and 92% ee. Notably, only cis-cyclobutane isomers were observed in these transformations, and the stereochemistry was preserved throughout the processes mentioned above.

Fig. 6: Late-stage transformations.
figure 6

aOnly cis-isomers were observed in these transformations. bReacting at 2 mmol scale from 3a.

Full size image

In summary, we have developed an asymmetric 1,2-carbon or boron migration mediated by ring-strained B-ate complexes, enabling the construction of atropisomers fearturing cis-cyclobutane boronic esters in good yields and excellent enantioselectivities. This methodology incorporates a diverse range of migratory functional groups, including aryl, vinyl, alkyl and boron groups. It is also compatible with natural product-derived organoborons, where the complex functional groups in the substrates have a negligible impact on the reaction. Additionally, mechanistic studies have been conducted to better understand the stereoselective control in the DYKATs. The transformations of the target products further demonstrate their practicality as valuable building blocks in synthetic chemistry.

Methods

General procedure for the preparation of chiral atropisomers featuring cyclobutane boronic esters

To a 10 mL flame-dried Schlenk-type sealed tube with a magnetic stir bar under N2 atmosphere was added 2 (0.4 mmol, 2.0 equiv.), p-methylphenylbicyclobutyl sulfoxide [1.1.0] (76.9 mg, 0.4 mmol, 2.0 equiv.) and THF (1.5 mL). The solution was stirred thoroughly and cooled to −78 °C, followed by the slow addition of tBuLi (0.4 mmol, 2.0 equiv., 1.3 M in pentane). The mixture was stirred for 5 minutes at −78 °C, then warmed to room temperature and stirred for an additional 5 minutes. After that, it was transferred to the glovebox. In the glovebox, Pd2(dba)3 (4.6 mg, 0.005 mmol, 2.5 mol%), (Ra)-L7 (8.8 mg, 0.012 mmol, 6 mol%) and toluene (0.4 mL) were added to a 2 mL vial containing a dried magnetic stir bar. The catalyst solution was stirred for 30 min at room temperature and then transferred into the reaction vial, followed by the addition of toluene (0.6 mL, used to rinse the Pd2(dba)3/(Ra)-L7 vial) and birayl triflate 3 (0.2 mmol, 1.0 equiv.). The reaction vial was sealed, removed from the glovebox, and heated to 60 °C, with the reaction progress monitored by TLC analysis. Upon completion, the reaction was quenched through 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 the solvent was removed under reduced pressure, the residue was purified by flash column chromatography on silica gel to afford the analytically pure products 4.