Introduction

Chirality is a fundamental characteristic of nature and covers various basic macromolecules, such as proteins, sugars, amino acids, and enzymes. Moreover, chiral molecules have a wide presence in natural products, drugs, and organic materials. Generally, the vast majority of chiral molecules contain one or more chiral elements, namely, central chirality, axis chirality, or/and planar chirality (Fig. 1a)1,2,3,4,5,6,7,8. In addition to these substructures, which are responsible for generating chirality, the entire molecular architecture can also exhibit chirality, as exemplified by helicenes and helicene-like compounds, which are helically twisted structures (Fig. 1a)9,10,11,12,13,14. In contrast, other chiral molecules without the aforementioned chiral elements have been largely overlooked. In 1994, the Böhmer group reported that calix[4]arenes with nonsymmetric substituents have chirality because the skeleton is deprived of symmetry. Unlike conventional central, axial, planar, or helical chirality, this unique chirality is defined as the concept of “inherent chirality” to decipher the intrinsic nature of molecules with concave and rigid conformations (Fig. 1a)15. In addition to calix[4]arenes16,17,18, inherent chirality has been discovered in some aromatic medium-sized rings with rigid nonplanar conformations19,20,21,22. For example, tetraphenylene derivatives, featuring an eight-membered polyarene composed of four alternating ortho-annulated benzenes, demonstrate a unique rigid saddle-shaped geometry and have inherent chirality with unsymmetrical substitution23,24,25,26,27. However, the application of asymmetric catalysis for the syntheses of these inherently chiral medium-sized cycles, especially those bearing a saddle-shaped eight-membered ring, is still in its infancy, and the corresponding literature documented thus far is quite limited. In 2009, Shibata and coworkers pioneered a Rh-catalyzed asymmetric cycloaddition of two triynes toward chiral eight-membered tetraphenylenes (Fig. 1b, left, 1st)28. In 2021, Luo, Zhu and coworkers reported elegant palladium-catalyzed double isocyanide insertion for the synthesis of chiral saddle-shaped eight-membered dibenzo[e,g][1,4]diazocin-6(5H)-ones (Fig. 1b, left, 2nd)29. Later, the same group accomplished the highly enantioselective synthesis of eight-membered dibenzo[e,g][1,4]diazocines through chiral phosphoric acid (CPA)-catalyzed formal (4 + 4) annulation between [1,1’-biphenyl]−2,2’-diamines and benzyls (Fig. 1b, middle)30. In 2023, the Yang group described impressive CPA-catalyzed kinetic resolution and dynamic kinetic resolution approaches for the asymmetric synthesis of inherently chiral 9,10-dihydrotribenzoazocines31 (Fig. 1b, right, 2nd). The same group subsequently achieved a CPA-catalyzed dimerization of 2-acylbenzoisocyanates toward inherently chiral saddle-shaped eight-membered [1,5]diazocines (Fig. 1b, right, 1st)32. Despite these limited advances, the development of innovative and general catalytic enantioselective strategies to access new families of chiral saddle-shaped eight-membered heterocycles is highly desirable but challenging, because of the unfavorable transannular interactions and entropic factors associated with constructing medium-sized rings.

Fig. 1: Background introduction and our strategy for synthesizing inherently chiral saddle-shaped eight-membered lactones.
figure 1

a Research status of the catalytic enantioselective synthesis of chemicals with different types of chirality. b (I) Strategies for the enantioselective synthesis of eight-membered rings. c Research advances in asymmetric transformations of alkynyl acyl azoliums. d Overview of this work.

Full size image

N-heterocyclic carbene (NHC) catalysis represents an appealing protocol for the construction of central or/and atropisomeric chiral compounds33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53, most of which involve the construction of axially chiral molecules through asymmetric (3 + 3) cycloaddition via the use of alkynyl acyl azolium as a C3 synthon54,55,56,57,58,59. To the best of our knowledge, no example of the enantioselective assembly of inherently chiral eight-membered lactones catalyzed by N-heterocyclic carbenes has been reported (Fig. 1c)60,61, which might arise from the following three issues: (1) the lack of appropriate and available precursors; (2) overcoming unfavorable transannular interactions and entropic penalties associated with their medium-sized ring formation; and (3) efficient control of enantioselectivity. To continue our interest in building medium-sized rings62,63,64 and achieve this goal, pro-axially chiral 1-(2-indolyl)naphthalen-2-ols 1 with 1,5-binucleophilic sites on biaryls were designed and synthesized (Fig. 1d). We speculated that the preformed 1-(2-indolyl)naphthalen-2-ols may react with chiral NHC-bounded alkynyl acyl azoliums as active intermediates generated from ynals54, enabling catalytic formal (5 + 3) annulation to furnish saddle-shaped eight-membered lactones with inherent chirality (Fig. 1d)

In this work, we show efficient and practical access to highly enantioenriched saddle-shaped eight-membered lactones with inherent chirality starting from readily available 1-(2-indolyl)naphthalen-2-ols with 1,3-dielectrophilic α,β‐alkynals through asymmetric oxidative NHC catalysis (Fig. 1d).

Results

Reaction optimization

To test our initial hypothesis, we began by studying the organocatalytic formal (5 + 3) annulation reaction of 1-(2-indolyl)naphthalen-2-ols with α,β‐alkynals as test substrates to develop asymmetrical access to highly enantioenriched eight-membered lactones. Diverse NHC pre-catalysts, bases and solvents were evaluated at 20 °C, as depicted in Table 1. Initially, the reaction of 1-(2-indolyl)naphthalen-2-ol 1a with 2a was conducted under established conditions adapted from the previously reported enantioselective construction of the C-N axes of thiazine derivatives57. With NHC precursor A with an N-mesityl group and Sc(OTf)3 as a Lewis acid catalyst, the reaction in THF in the presence of 3,3’,5,5’-tetra-tert-butyldiphenylquinone (DQ) as an oxidant and 4-dimethylaminopyridine (DMAP) as a base gave desired product 3a, albeit with low yield (10%) and enantioselectivity (59% ee, entry 1). Preliminary results prompted us to further optimize the reaction conditions. Several other NHC precursors with an N-perfluorophenyl B, an N−2,6-diethylphenyl C or an N-phenyl D substituent were then examined (entries 2-4), demonstrating that all these catalysts displayed higher catalytic performances in terms of enantioselectivity than did A; moreover, to our delight, the latter precursor D proved to be the better choice for this asymmetric transformation, which furnished 45% yield and 90% ee (entry 4). The bases had important effects on the reaction efficiency and enantioselectivity (entries 5 − 8). The use of Et3N completely suppressed the generation of 3a (entry 5), whereas DABCO resulted in a slightly increased yield but a significant decrease in the ee value compared with that of DMAP (entry 6). Exchanging organic bases with inorganic bases, such as K2CO3 and K3PO4, did not improve either the chemical yield or the enantioselectivity (entries 7-8). The yields and ee values of product 3a associated with the NHC precursor D and DMAP in several aprotic solvents were subsequently summarized as follows (entries 9 − 12): toluene (17%, 75% ee), CH3CN (35%, 70% ee), 1,4-dioxane (22%, 81% ee), and DCE (31%, 87% ee). The above observations indicated that all these solvents were inferior to THF in terms of yield and ee value 3a. An increase in the amount of DQ to 3.0 equivalents is beneficial for ensuring the yield of 3a while maintaining enantioselectivity (entry 13). Furthermore, increasing the concentration of substrate slightly improved the enantioselectivity (entry 14). To our delight, the reaction was more efficient, delivering better yield (75%) and higher enantioselectivity (97% ee) while decreasing the catalyst loading to 15 mol% (entry 15). Furthermore, a decrease in the amount of DMAP slightly improved the yield (entry 16). Finally, slightly increasing the amount of substrate 2a resulted in 93% yield without loss of enantioselectivity (entry 17).

Table 1 Optimization of reaction conditions.aView full size image
Full size table

Evaluation of the substrate scope

Using the optimized conditions, the generality of the NHC-catalyzed formal (5 + 3) annulation reaction with respect to an array of 1-(2-indolyl)naphthalen-2-ols with α,β‐alkynals was examined (Fig. 2). To our delight, a number of 1-(2-indolyl)naphthalen-2-ols with different steric and electronic properties in both the indole and naphthalene rings were compatible, as shown in Fig. 2. For the indole ring, different substituents, such as fluoro (1b), chloro (1c and 1 h), bromo (1d, 1i and 1j, 1k), trifluoromethyl (1e), ester (1 f), and methyl (1 g), at the C4 to C7 positions were well tolerated in this reaction, furnishing the corresponding eight-membered lactones 4 − 13 in moderate to good yields (41%–90%) and excellent enantioselectivities (92%–97% ee). Of these groups, an obvious steric effect was observed because the C4 bromo analogue 1k, which has strong steric hindrance, demonstrated significantly decreased enantioselectivity (13, 81% ee). Interestingly, even in a challenging case where the trifluoromethyl functional group was a strong electron-withdrawing group residing at the C5 position, this catalytic protocol was applicable, as product 7 was generated in 74% yield and 92% ee. For the naphthalene ring, either electronically rich (e.g., methyl, ethyl, isopropyl and phenyl) or deficient (e.g., bromo) substituents at different positions proceeded smoothly through organocatalytic formal (5 + 3) annulations with comparable efficiencies, accessing desired products 14 − 20 in 52-75% yields and 92-98% ee values.

Fig. 2: Scope of 1-(2-indolyl)naphthalen-2-ols.
figure 2

Reaction conditions: 1 (0.3 mmol), 2a (1.05 mmol), NHC pre-catalyst D (15 mol%), Sc(OTf)3 (20 mol%), DMAP (0.3 mmol), THF (7.5 mL), and 4 Å MS (300 mg) at 20 oC, for 3 days; the yield refers to the isolated yield based on 1; ee values were determined by HPLC.

Full size image

The scope of the α,β‐alkynals was proven to be quite broad (Fig. 3). Electronically rich and poor α,β‐alkynals were compatible with the reaction conditions, and products 21 − 28, which possessed para-, meta- and ortho-methyl, para-ethyl, ortho-fluoro, para-chloro, meta-bromo and para-nitro substituents, were all synthesized in acceptable yields and high enantioselectivities. Among them, incorporating ortho-methyl and para-nitro groups of the α,β‐alkynals obviously decreased the enantioselectivity of the annulation reaction, and eight-membered lactone products 23 and 28 were formed with 80% and 84% ee, respectively. In addition to phenyl-substituted α,β‐alkynal derivatives, heteroaryl α,β‐alkynals such as 2-thienyl and 3-indolyl could also participate in this transformation, and compounds 29-45 were produced. Notably, a brief survey of the potential variations in the indole ring of α,β‐alkynals 2 was conducted. The effect of the electronic properties and positions of the substituents on the indole ring in this transformation was investigated. Various functional groups, such as methyl (C4, 2 l; C5, 2 m; C7, 2n), methoxy (C6, 2o), fluoro (C5, 2p), chloro (C4, 2q; C5, 2r; C6, 2 s), bromo (C5, 2t), and nitro (C5, 2 u) groups at different positions, were examined to demonstrate the compatibility of these asymmetrical annulations, and all of them were converted into the corresponding products 32-40. These results indicated that the position of the substituents on the indole ring has a profound impact on the enantioselectivity. Substrates bearing C4 substituents on the indole ring, probably due to steric hindrance, were reluctant to undergo this annulative process, as remarkable decreases in both chemical yields and enantioselectivities were observed (31 and 36). The variations in 1-(2-indolyl)naphthalen-2-ols were then investigated by combining 3-indole-substituted α,β‐alkynal 2k. Both electron-donating (methyl) and electron-withdrawing (chloro and bromo) groups in the indole or naphthalene ring were applicable, and compounds 41-45 were formed with ≥90% ee. Swapping the aryl moiety of α,β‐alkynals with alkyl groups led to the formation of products 46-49 in moderate to excellent yields and high enantioselectivities. Several representative long-chain alkyl groups containing n-butyl, n-amyl, ether and TBS-protected alcohols all showed good reactivity profiles, producing corresponding products 46-49 with 91%–94% ee values.

Fig. 3: Scope of α,β‐alkynals 2.
figure 3

Reaction conditions: 1 (0.3 mmol), 2 (1.05 mmol), NHC pre-catalyst D (15 mol%), Sc(OTf)3 (20 mol%), DMAP (0.3 mmol), THF (7.5 mL), and 4 Å MS (300 mg) at 20 oC, for 3 days; the yield refers to the isolated yield based on 1; ee values were determined by HPLC.

Full size image

Study of product stability and synthetic applications

To further investigate the conformational stability of eight-membered lactones, racemization experiments were performed (Fig. 4). Initially, the newly formed 3 was heated to 80 °C in isopropanol, and the ee decreased by 15.4% after 4 h. A racemization half-life of 5.5 × 104 h (at 25 °C) and a racemization barrier of 29.4 kcal/mol were calculated from these results, indicating the configurational stability of eight-membered lactones at relatively low temperatures. To verify the accuracy of the racemization barrier test, we further studied the racemization process of 3 by DFT calculations. As shown in Fig. 4 and Supplementary Fig. 20, the energy barrier of racemization is 30.7 kcal/mol, which supports the experimental results. Moreover, we changed the naphthalene ring to an ortho-substituted phenyl ring, such as ortho-methyl or ortho-chloro, and computationally investigated the possibility of synthesizing eight-membered lactones 50 and 51. The computational results showed that the energy barriers to enantiomerization were 24.0 kcal/mol and 23.6 kcal/mol, respectively (see Supplementary Fig. 20), implying that the relatively lower calculated barrier to enantiomerization for these two compounds revealed their configurational lability and that the enantioselectivity was considered uncontrollable. To confirm this hypothesis, reactions of 52 and 53 were independently conducted under standard conditions, delivering products 50 and 51 in 73% and 81% yields, respectively, but without observation of enantioselectivity, which was in good agreement with the computational results.

Fig. 4: Study on product stability.
figure 4

a Experimental and DFT calculated determination of the enantiomerization barrier of 3. b DFT calculations of the energy barriers of the enantiomerization of 50 and 51. c Experimental validation.

Full size image

A scale-up experiment using 1a (1.0 mmol) with 2a under standard conditions afforded 3 in a reduced yield (73%) without an obvious decrease in enantioselectivity (97% ee, Fig. 5a). The potential synthetic utility of the current developed protocol for accessing other chiral molecules was further highlighted by late-stage transformations. N-acetylation of 3 was carried out in the presence of DMAP and Et3N to produce N-acetyl product 54 with a remarkable 98% ee by using Ac2O as an acetylation reagent (Fig. 5b)65. Similarly, the use of Bc2O led to the formation of N-Boc product 55 without a loss of enantioselectivity66. Next, ring opening of the oxocin-2-one framework occurred readily in the presence of different esterification reagents, affording axially chiral biaryls 56 and 57 without loss of enantioselectivity. Owing to the presence of two chiral axes in these two compounds, we performed DFT studies to calculate the rotational barriers of these chiral axes and determine whether they are biaxial chiral compounds. DFT calculations for compounds 56 demonstrate that the rotation barriers of C(indole)-C(Np) and C(indole)-C(alkene) axes are 37.7 and 21.4 kcal/mol, respectively, indicating that compound 56 has only one rotationally hindered chiral axis. A similar phenomenon was observed for compounds 57 and 59, in which the high rotation barriers of C(indole)-C(Np) axis were calculated (57, 52.3 kcal/mol; 59, 41.5 kcal/mol) and associated with the low rotation barriers of C(indole)-C(alkene) axis (57, 21.2 kcal/mol; 59, 21.9 kcal/mol). The presence of lithium ethoxide in Et2O resulted in axially chiral biaryls 56, whereas compound 57 was afforded by using methanol as the esterification reagent and Et3N as the base (Fig. 5b). In addition, treatment of 3 with methyl iodide gave N-methyl product 58 with significantly decreased enantioselectivity (82% ee). The reason may be that the free N-H bond of product 3 loses a proton in the presence of K2CO3, decreasing the steric hindrance and causing racemization. After simple recrystallization, the ee of N-methyl product 58 improved to 99%, and this was followed by ring-opening with methanol and Et3N to access axially chiral biaryls 59 in 95% yield and 99% ee. Interestingly, de-esterification of 59 proceeded smoothly in the presence of diisopropylethylamine (DIPEA), triflic anhydride (Tf2O) and triflic acid (TfOH), affording terminal olefin 60 in 57% yield and 99% ee. The reaction of 3 with chlorodiphenylphosphine furnished monophosphine product 61 in 89% yield and 95% ee67. The synthesized monophosphine product 61 was subsequently used as a chiral ligand for the asymmetric Pd-catalyzed allylic substitution of dimethyl malonate and the Tsuji-Trost reaction, and the desired product 62 was obtained in 85% yield but with very poor enantioselectivity (Fig. 5c)67.

Fig. 5: Scale-up reaction and synthetic transformations.
figure 5

a Scale-up experiment. b Chemical transformations of compound 3. c Application of compound 61. The yields are the isolated yields after purification by column chromatography. The ee values were determined by HPLC. The Z/E ratios were determined by 1H MNR.

Full size image

Possible reaction pathways

To gain mechanistic insights into this process, several control experiments were conducted. 1a reacted with 2a under standard conditions without Sc(OTf)3, and product 3 was afforded in both decreased yield and enantioselectivity (81%, 80% ee), showing that Sc(OTf)3 had a positive effect on the efficiency of this transformation in terms of yield and ee (Fig. 6a). This interesting observation enabled us to further understand this process. From these results, we reasoned that Sc3+ may act as a linker to fix alkynyl acyl azolium I and naphthalen-2-olate III, which enables the intermolecular nucleophilic addition pathway in the absence of scandium to occur in an intramolecular pattern and thereby facilitates the Michael addition of indole to alkynyl acyl azolium (Fig. 7), thus leading to improvements in both the yield and the enantioselectivity. Without DQ or the NHC catalyst, the reaction did not proceed, and the starting materials were recovered (Fig. 6b and c), suggesting that alkynyl acyl azolium may be a key intermediate and that the NHC catalyst is crucial for this transformation. To confirm the role of the free N-H bond in substrate 1, N-methyl-protected 1-(2-indolyl)naphthalen-2-ol 63 was reacted with 2a in the absence of Sc(OTf)3 under standard conditions, furnishing corresponding product 64 in 63% yield but without observation of the ee (Fig. 6d). The above reaction did not work without an NHC catalyst (Fig. 6e). These results implied that the substituent at the N-atom of 1-(2-indolyl)naphthalen-2-ol severely affects the enantioselective control of this transformation.

Fig. 6: Investigation of the mechanism.
figure 6

a Control reaction of 1a with 2a without Sc(OTf)3. b Control reaction of 1a with 2a without DQ. c Control reaction of 1a with 2a without Cat-D. d Control reaction of 63 with 2a without Cat-D and Sc(OTf)3. e Control reaction of 63 with 2a without Sc(OTf)3.

Full size image
Fig. 7
figure 7

Proposed reaction pathway.

Full size image

Based on the above experimental results and previous reports54,55,56,57,58,59, a proposed mechanism is depicted in Fig. 7. With 1-(2-indolyl)naphthalen-2-ol 1a and 3-phenylpropiolaldehyde 2a as representative examples, this process might initially involve in-situ-generation of a free NHC catalyst from precatalyst D under basic conditions, and the subsequent addition of the NHC catalyst to 2a gives Breslow intermediate I. Then, intermediate I is oxidized by DQ, followed by coordination of Sc(OTf)3 to yield the alkynyl acyl azolium-[Sc]-complex II. Intermediate II captures naphthalen-2-olate III, derived from 1a under basic conditions, by coordination, affording [Sc]-complex IV. Michael addition of indole to alkynyl acyl azolium yields allenolate intermediate V, which undergoes proton transfer (PT) and subsequent lactonization to afford product 3 and simultaneously regenerates the NHC catalyst and [Sc]-complex for the next catalytic cycles.

Discussion

In summary, we developed an oxidative NHC catalytic formal (5 + 3) annulation strategy starting from 1-(2-indolyl)naphthalen-2-ol and α,β‐alkynals with the aid of Sc(OTf)3, and used this process to produce a wide range of challenging inherently chiral saddle-shaped eight-membered lactones with structural diversity in moderate to good yields and high enantioselectivities. The stabilities of the inherently chiral eight-membered lactones were evaluated via both experimental and computational methods, which demonstrated that these molecules are quite stable at relatively low temperatures. Both the indole and oxocin-2-one scaffolds in our eight-membered rings are commonly used structural motifs in bioactive and other functional molecules and can be easily derivatized into a series of functionally diverse chiral molecules, in which preliminary deconstructive esterification of the oxocin-2-one ring to axially chiral biaryls was showcased. We believe that our developed methodology will inspire more investigations on inherently chiral medium-ring skeletons, especially those bearing eight-membered rings. Further studies on the enantioselective synthesis of inherently chiral medium-ring-embedded molecules, especially those with potential applications in pharmaceutical and medical chemistry, are currently underway in our laboratories.

Methods

General procedure for the NHC-catalyzed domino reaction of 1-(2-indolyl)naphthalen-2-ols and α,β‐alkynals. Chiral NHC pre-catalyst D (0.045 mmol, 15 mol%, 17 mg), DMAP (0.3 mmol, 1.0 equiv, 37 mg), Sc(OTf)3 (0.06 mmol, 20 mol%, 30 mg), 4 Å molecular sieves (300 mg), DQ (0.9 mmol, 3.0 equiv, 368 mg) and substituted 1-(2-indolyl)naphthalen-2-ols 1 (0.3 mmol) was added to a 20 mL vial equipped with a magnetic stir bar. After that, THF (7.5 mL) and α,β‐alkynals 2 (1.05 mmol, 3.5 equiv) were added, and the reaction mixture was allowed to stir for 3 days at 20 °C. After consumption of 1-(2-indolyl)naphthalen-2-ols 1(monitored by TLC), the reaction mixture was directly subjected to column chromatography on silica gel (eluent, petroleum ether/ethyl acetate = 20:1 V/V) to afford desired products 3-49.