Introduction

Novel catalytic and asymmetric reactions are being developed continuously, which compensate for the drawbacks in previous methods and make new chiral molecules easily accessible1,2,3,4. In particular, for the control of a stereogenic center, several catalytic and reliable methods have been developed with the least required functional group in a substrate, owing to the accumulation of numerous valuable precedents5,6,7. However, controlling several types of chiralities, such as axial8,9,10, helical11,12, and planar chiralities13,14, remains difficult; therefore, even a minor modification in a substrate frequently causes a substantial loss of yields or selectivities15,16. Thus, even though newly developed individual methods represent the cutting-edge of asymmetric catalysis, a versatile strategy that can control a difficult stereogenic element with a high degree of generality would be ideal and desirable15,16.

In well-established asymmetric reactions, secondary interactions between asymmetric catalysts and substrates frequently play a significant role in achieving high enantioselectivity6,7,17,18. For example, in the Sharpless asymmetric epoxidation, a titanium-coordinated hydroxyl group close to an alkene confers facial selectivity (Fig. 1a, top)19. In addition, β-ketoesters are recognized as the optimal substrate for the Noyori asymmetric hydrogenation, because of the interaction between the metal and the carbonyl oxygen (Fig. 1a, bottom)20. These illustrative examples suggest that introducing a novel and flexible secondary interaction would create a new path, particularly in the challenging topics of asymmetric catalysis. In this regard, the present study proposes a general strategy for atroposelective dynamic kinetic resolution21,22,23,24,25,26,27,28,29,30,31,32,33,34 in which one aromatic ring is substituted by the aid of a catalytic chiral phosphoric acid35,36,37 while the other aromatic ring forms a secondary interaction with the catalyst38,39,40 (Fig. 1b).

Fig. 1: Development of a general strategy for catalytic atroposelective synthesis.
figure 1

a Examples of key secondary interactions in asymmetric catalysis. b General concept. c Atropisomerism in arylindoles. d Examples of atropisomeric arylindoles. e Catalytic atroposelective synthesis of substituted arylindoles. f This work.

Full size image

Highly substituted indoles play a crucial role in organic chemistry as essential structural motifs in natural products41, biologically active compounds42,43, and material sciences44. Due to the higher electron density of a nitrogen-containing five-membered ring (A ring in Fig. 1c) compared to a carbocyclic ring (B ring in Fig. 1c)45, substitutions are typically concentrated at the N−1, C−2, and C−3 positions. Specifically, the introduction of an aromatic ring at these positions has significantly elongated the indole scaffold both in nature and in the laboratory; consequently, the discovery of atropisomerism in indoles continues to increase (Fig. 1d)46,47,48. Due to its biological49 and synthetic significance50, numerous efforts have been directed to the catalytic control of a stereogenic axis in indoles, which represents the state of the art in asymmetric catalysis51,52,53,54,55,56,57. Among them, catalytic dynamic kinetic resolutions of 3-aryl-1H-indoles have been investigated because of the high reactivity at the C−2 position (Fig. 1e). For example, the Shi group reported an atroposelective intermolecular functionalization of 3-arylindoles with multiple electrophiles catalyzed by chiral phosphoric acid54,55,56. More recently, the Fu group employed indole-3-one as an electrophile to control the stereogenic center and axis57. They hypothesized that the interaction between the N−H in the indole and the P=O in chiral phosphoric acid is essential for achieving high enantioselectivity38,39,40. While these findings have provided general and efficient for synthesizing atropisomeric 3-arylindoles, they have intrinsic limitations for other arylindoles that lack interaction, such as N-arylindole, 2-arylindole, and N-substituted-3-arylindole (Fig. 1e).

Our group recently reported the chiral phosphoric acid−catalyzed atroposelective Pictet−Spengler cyclization of N-arylindoles and biphenyls58,59. In this reaction, an N−H group was introduced to a phenyl ring, creating a critical interaction with the catalyst. With our ongoing research interest in catalytic and atroposelective reactions, we hypothesized that intermolecular atroposelective functionalization would be realized by the introduction of an N−H group into a substrate. Furthermore, o-amino/o-amidobiaryl substrates could be extensively utilized as a platform for catalytic and atroposelective dynamic kinetic resolution. To illustrate our theory, we envisioned the atroposelective synthesis of not only N-arylindoles, but also 2-arylindoles and 3-arylindoles. To the best of our knowledge, one catalytic method that can control an axis at three separate positions of indoles has not been developed. Moreover, atroposelective synthesis of 2-arylindole via dynamic kinetic resolution is rare. Herein, we provide a highly atroposelective dynamic kinetic resolution of three distinct types of arylindoles with ketomalonates catalyzed by a chiral phosphoric acid, thereby providing an effective method for indoles with a high degree of generality (Fig. 1f).

Results

Catalytic dynamic kinetic resolution of N-/2-/3-arylindoles

Generally, atropisomerism occurs when at least three substituents surround a stereogenic single bond60,61. To demonstrate our general concept shown in Fig. 1b, N-arylindole, 2-arylindole, and 3-arylindole containing a benzamide group at the ortho position of the phenyl rings were designed. We hypothesized that the catalytic functionalization at the C−2 position of N-arylindole and 3-arylindole would create a sufficient rotational barrier to maintain stereo-configuration. However, because the remaining N−1 and C−3 positions are more reactive than the C−2 position, the methyl group was introduced to the C−3 position of N-arylindole (1a) and N−1 position of 3-arylindole (6a). After the catalytic reaction at the C−3 position, we anticipated that a methyl group would be required at the N−1 position of 2-arylindole (4a) to achieve a high rotational barrier. We believe that these three regioisomeric substrates (1a, 4a, and 6a) demonstrate the broad applicability of our technique.

To test our hypothesis, chiral phosphoric acids were screened (Fig. 2) in chloroform at 40 °C with N-arylindole (1a) and diethyl ketomalonate (2a) (For the detailed electrophile screening, see the Supplementary Information). Initially, we examined chiral phosphoric acids (P1) substituted with 4-methoxyphenyl groups, which smoothly induced the reaction to produce the desired product (3aa) with a 56% yield despite the absence of enantioselectivity (Fig. 2, entry 1–1). The replacement of the 4-methoxy group with the 4-tert-butyl group did not enhance enantioselectivity (Fig. 2, entry 1–2). The reaction of 1a with the 3,5-diphenyl-substituted catalyst (P3) exhibited marginally increased enantioselectivity (26% ee), indicating that the number and position of substituents play a crucial role in determining selectivity (Fig. 2, entry 1–3). Interestingly, the catalysts (P4 and P5) bearing electron-withdrawing groups produced the opposite atropisomer to that obtained with P3 (Fig. 2, entries 1–4 and 1–5), where the 4-trifluoromethyl substituted catalyst (P4) performed better than the 3,5-bistrifluoromethyl substituted catalyst (P5) (20% ee vs. 3% ee). These results suggested that these catalysts will operate in a manner that is different from that of P1−P3. To achieve high enantioselectivity, catalysts with bulkier substituents, such as 9-phenanthryl (P6), 9-anthracene (P7), and 2,4,6-isopropylphenyl (P8), were utilized. In these reactions, the greater the size of the substituent group was, the greater enantioselectivity was observed (P6, 55% ee; P7, 59% ee; P8, 69% ee). To our delight, higher enantioselectivity was achieved with the 2,4,6-tricyclohexyl-substituted catalyst (P9), which resulted in 86% yield and 82% ee of the desired product (Fig. 2, entry 1–9). When the VAPOL catalyst (P10) was employed, almost no enantioselectivity was observed (Fig. 2, entry 1–10). Using the most effective catalyst (P9), we screened several solvents (See the Supplementary Information for details) and determined that toluene was the optimal solvent in terms of yield and enantioselectivity (99% yield and 89% ee). While excellent enantioselectivity (95% ee) was observed at lower temperatures (–5 °C), chemical yield decreased to 74% (Fig. 2, entry 1–12). Fortunately, we observed that the reduced yield was compensated for by the addition of 3 Å molecular sieves, resulting in 99% yield and 96% ee of the required product (Fig. 2, entry 1–13).

Fig. 2: Selected optimizations of the reaction conditions.
figure 2

aIsolated yields. bEnantiomeric excesses were determined via chiral high-performance liquid chromatography analysis. cThe values in parentheses represent the enantiomeric excesses of the opposite enantiomer.

Full size image

After establishing the optimal reaction conditions for N-arylindoles, we turned our attention to 2-arylindoles. The reaction of the regioisomeric 2-arylindole (4a) was performed using the previously established optimal reaction conditions that quickly yielded the desired product (5aa) in 99% yield and 66% ee (Fig. 2, entry 2–1). Due to the increased reactivity of the C−3 position compared to the C−2 position, the reaction occurred even in the absence of a catalyst (18% yield). To decrease the background rate, a lower concentration of 5aa was shown to be beneficial, displaying 72% ee (Fig. 2, entry 2–2). In addition, a lower temperature (−20 °C) resulted in greater enantioselectivity (81% ee) despite the longer reaction time (Fig. 2, entry 2–3). The highest enantioselectivity for 2-arylindole was achieved at −78 °C, providing 5aa in 99% yield and 83% ee (Fig. 2, entry 2–5).

As 3-arylindole is the reversed form of N-arylindole, we assumed that 3-arylindole would also exhibit a high degree of selectivity. In fact, when we performed the reaction of 3-arylindole (6a) under optimal conditions for N-arylindoles, we observed an enantioselectivity of 86% ee (Fig. 2, entry 3–1). However, in this reaction, the chemical yield drastically decreased, yielding only 7% of the required product (7aa). To increase the reaction rate while maintaining enantioselectivity, the reaction parameters were evaluated. When Lewis acids were employed to accelerates the reaction rate, slightly higher yields but lower enantioselectivities were observed (See the Supplementary Information for details). The reaction in dichloromethane rather than toluene produced a slightly better yield (Fig. 2, entry 3–2). As the optimal concentration corresponded to the minimum solvent volume (0.2 mL) required to properly mix the reaction, a higher concentration could not be used. When the reaction was conducted at a higher temperature, enantioselectivity decreased significantly. Eventually, the excess quantity of 2a was used, providing 7aa in 39% yield and 89% ee (Fig. 2, entry 3–3). A further increase in catalyst loading increased the yield to 65% while retaining enantioselectivity (Fig. 2, entry 3–5). We believe that these optimizations reflect the different reactivities of the C−2 and C−3 positions in the indole and a subtle distinction between substrates in asymmetric catalysis.

Evaluation of the substrate scopes

Using the optimal reaction conditions for N-arylindoles, 2-arylindoles, and 3-arylindoles, the substrate scopes were explored. Initially, substituted N-arylindoles were investigated, which demonstrated the broad applicability of our methodology (Fig. 3). When the bottom aromatic ring was substituted with a methyl group at the ortho position of the NHBz group, excellent enantioselectivity was observed; however, the chemical yield was low (45% yield, 96% ee). To address this issue, 20 mol% of P9 was utilized, resulting in 99% yield and 96% ee for 3ba. The methyl substitution at the para position relative to the stereogenic axis was found to be acceptable and resulted in excellent enantioselectivity (76% yield, 95% ee). The reaction of the methoxy-substituted substrate at the para position of NHBz (1d) produced the intended product (3da) with even higher enantioselectivity but with a moderate yield (52% yield, 99% ee). The chemical yield was easily enhanced with 20 mol% of P9 while maintaining enantioselectivity (93% yield, 99% ee). However, the substrate (1e) bearing a methoxy group in the ortho position of the stereogenic axis did not activate the reaction under the optimal reaction conditions, likely due to the steric repulsion between the methoxy group and 2a. The replacement of a benzamide group with a benzylamino group and a methanesulfonamide group produced the alcohol products 3fa (81% yield and 99% ee) and 3ga (99% yield and 96% ee). Accordingly, the reaction of 3h with a benzylamino group and a fluoride group produced a high atroposelectivity product (80% yield and 95% ee).

Fig. 3: Catalytic atroposelective synthesis of N-arylindoles.
figure 3

Unless otherwise noted, the reactions were conducted with 1 (0.050 mmol, 1.0 equiv), 2 (0.075 mmol, 1.5 equiv), P9 (0.005 mmol, 10 mol%, 5.0 mg), and 3 Å molecular sieves (40 mg) in PhMe (0.2 mL) at −5 °C. a20 mol% of P9 was employed. bThe reaction was performed at room temperature with 20 mol% of P9.

Full size image

A benzamide group was replaced with either an alkyl or aryl group to achieve high enantioselectivity. In fact, the methyl-substituted substrate (1i) demonstrated a considerable degree of enantioselectivity (84% ee). The N-naphthylindole (1j) reaction produced the intended product in 71% yield and 27% ee. However, the substrate (1k) bearing an isopropyl group did not affect the reaction. While the modification of the indole ring was comparable with our methodology, the reactivity varied depending on the substituent. For example, 5-methoxy-N-arylindole (1l) and 5-chloro-N-arylindole (1m) exhibited significantly different reactivity but comparable enantioselectivity (3la, 93% yield, 95% ee; 3ma, 34% yield, 97% ee). The chemical yield of 3ma was increased to 70% by utilizing 20 mol% of P9. Further substitutions of 1l with electron-withdrawing groups, such as cyano and fluoride groups, were tolerated even though 20 mol% of P9 was necessary to obtain a high yield of 3na. The methyl group at the C−3 position was substituted with a phenyl group or a benzyl group (1p and 1q), which did not yield the desired products. This incompatibility can be attributed to the steric hindrance between the catalyst and the C−3 position of indole in the transition states. When the reaction was performed with dimethyl ketomalonate (2b) or diisopropyl ketomalonate (2c) instead of diethyl ketomalonate (2a), the expected products (3ab and 3ac) were obtained in 73% yield, 91% ee and 98% yield, 88% ee, respectively. As unsymmetric electrophiles might produce and regulate the stereogenic axis and quaternary stereogenic center, ethyl trifluoropyruvate (2d) was used as an electrophile for the catalytic reaction. Even though the reaction was slow under optimal conditions, excellent enantio- and diastereoselectivities were observed (17% yield, >20:1 dr, and 96% ee). When the reaction was performed with 20 mol% of P9 at room temperature, the desired product (3ad) was obtained in 50% yield, >20:1 dr, and 97% ee. The reaction of 1a with methyl trifluoropyruvate (2e) yielded the desired product (3ae) in 6% yield, >20:1 dr, 92% ee under the optimal reaction conditions, and 49% yield, >20:1 dr, 89% ee with 20 mol% of P9 at room temperature. While methyl 2-oxoacetate (2f) produced the desired product albeit low selectivity (~1:1 dr, 73% ee), methyl 2-oxopropanoate (2g) and methyl 2-oxo-2-phenylacetate (2h) were found to be incompatible with our methodology. When the substrate with a 5-methoxy substitution (1l) was employed instead of 1a, the reaction rate increased, but the selectivity was either maintained or slightly lower. The reaction of 5-methoxy-N-arylindole (1l) with 2c produced 3lc with moderate enantioselectivity. The reaction of 1l with 2d, 2e, and 2f provided the desired product in higher yields and similar selectivities under the optimized reaction conditions. However, 2g and 2h still did not yield the desired product.

Substrate scopes for 2-arylindoles were then examined with the optimized reaction conditions, which overall exhibited high enantioselectivities (Fig. 4, left). When the methoxy group is introduced at the para position of the benzamide, excellent yield and selectivity were observed (5ba, 95% yield, 96% ee). The replacements of the benzamide group with 4-methylbenzamide and 4-bromobenzamide provided the desired products 5ca and 5da (5ca, 99% yield, 87% ee; 5da, 99% yield, 82% ee). The substrate (4e) containing a benzylamino group instead of a benzamide group was reacted with 2a to produce the desired product in 99% yield and 90% ee. When the benzamide group was substituted with an alkyl or aryl group, 2-arylindoles produced superior results than N-arylindoles (3ia, 3ja, and 3ka vs. 5fa, 5ga, and 5ha). For example, the reaction of a substrate containing a methyl group produced 5fa in 96% yield and 87% ee. The catalyst induced the reaction of 2-naphthylindole (4g) to produce 5ga in 40% yield and 77% ee. N-arylindole bearing an isopropyl group did not react with 2a; however, 2-arylindole substrate (4h) and 2a provided the desired product (5ha) with 59% yield and 43% ee, owing to the strong reactivity of the C−3 position. However, the reaction of the substrate containing a phenyl group (4i) failed to produce the desired product. 5-methoxy-2-arylindole (4j) interacted with 2a in the presence of P9 to produce 99% yield and 94% ee. Interestingly, a distinct tendency was observed compared to N-arylindoles when 2b and 2c were employed. The reaction between compound 4a and dimethyl ketomalonate (2b) produced the atropisomeric product with very low enantioselectivity (14% ee). However, diisopropyl ketomalonate (2c) exhibited an even higher degree of enantioselectivity than diethyl ketomalonate (2b) (96% ee vs. 83% ee), implying that the origin of enantioselectivity for 2-arylindoles would be markedly distinct from that of N-arylindoles. When the methyl group at the N−1 position was substituted with an isopropyl group (4k), the selectivity reduced significantly to 44% ee. The reaction of 4a with unsymmetrical ketones (1d and 1e) resulted in the desired product with excellent yield, but low selectivity.

Fig. 4: Catalytic atroposelective synthesis of 2- and 3-arylindoles.
figure 4

The reactions for 5 were conducted with 4 (0.050 mmol, 1.0 equiv), 2 (0.075 mmol, 1.5 equiv), P9 (0.005 mmol, 10 mol%, 5.0 mg), and 3 Å molecular sieves (400 mg) in PhMe (2.0 mL) at −78 °C. Unless otherwise noted, the reactions for 7 were conducted with 6 (0.050 mmol, 1.0 equiv), 2 (1.125 mmol, 15 equiv), P9 (0.020 mmol, 40 mol%, 20.0 mg), and 3 Å molecular sieves (40 mg) in PhMe (0.2 mL) at −5 °C. aThe reaction was performed in PhMe with 10 mol% of P9.

Full size image

Then, we investigated 3-arylindoles under optimal reaction conditions (Fig. 4, right). The addition of a methyl group to the para position of the stereogenic axis was tolerated (7ba, 87% yield, 87% ee). When the benzamide group was substituted with a methyl group (6c), yield and selectivity decreased, further emphasizing the importance of the benzamide group. Interestingly, substituting the 5-position of indole with a methoxy group increased reactivity as the catalyst induced the catalytic reaction gradually under the optimal reaction conditions for N-arylindole (7da, 99% yield, 86% ee; 7ea, 72% yield, 80% ee). While the substrates in which the methyl group at the N−1 position was substituted with an ethyl group (6f) or propyl group (6g) produced the desired products, the enantioselectivities were significantly reduced (40% ee for 7fa and 22% ee for 7ga). Substitution with a benzyl group (6h) or the removal of the substitution (6l) failed to yield the desired product. The reaction of 5-methoxy-3-arylindole (6d) with diisopropyl ketomalonate (2c) resulted in the desired product with 80% yield and 79% ee. The unsymmetrical electrophiles (2d and 2e) were employed for the reaction of 6d, which led to excellent yields and good to moderate enantioselectivities (>20:1 dr, 86% ee for 7dd and 9:1 dr, 77% ee for 7de).

Computational studies and mechanistic consideration

To investigate the configurational stability of reaction products, the rotational barriers of 3aa, 5aa, and 7aa were calculated, as shown in Fig. 5a. According to density functional theory (DFT) calculations (See the Supplementary Information for details)62, 3aa and 7aa would have similarly high rotational barriers (36.7 and 36.6 kcal/mol, respectively), whereas 5aa would have the highest rotational barrier (39.1 kcal/mol). These results indicated that the products would be configurationally stable and widely applicable in various areas without modification. The absolute stereochemistry was then determined by obtaining single-crystal X-ray structures of 3ba, 3ld, and 7da, which were assigned as axial R-configured (Fig. 5b). To clarify the situation and assign the absolute stereochemistry, a chiroptical method was employed63,64. The experimental electronic circular dichroism (ECD) spectra of 3aa, 5aa, and 7aa were measured and compared with their respective time-dependent DFT-calculated ECD spectra (See the Supplementary Information for details)62, as depicted in Fig. 5c. The measured ECD spectra of 3aa and 7aa well-matched the calculated ECD spectrum of axial R-configured molecules. These results demonstrated that our ECD calculations were in good agreement with the single-crystal X-ray structures. Similarly, the absolute stereochemistry of 5aa was determined to be axially S-configured.

Fig. 5: Computational studies and mechanistic consideration.
figure 5

a Calculated rotational barriers for 3aa, 5aa, and 7aa. b X-ray crystal structures of 3ba, 3ld, and 7da. c Calculated and experimental ECD spectra of 3aa, 5aa, and 7aa. d Transition state structures for synthesizing 3aa, 5aa, and 7aa. (All H atoms are omitted for clarity).

Full size image

To investigate the source of the observed enantioselectivity, computational studies were conducted (See the Supplementary Information for details) (Fig. 5d)62. Because of the voluminous substituents of chiral phosphoric acid (P9), the two-layer quantum mechanical (QM)/semiempirical (SE) ONIOM model was used65,66,67. The phosphoric acid group [–O2P(O)OH] in P9 and substrates were assigned to the QM layer while the remaining atoms of P9 were coated with the SE layer68. The QM layer was treated with M06-2X/6-31 G(d) while the SE layer was treated with PM6 for the geometry optimizations. Single-point energies of these optimized structures were calculated using M06-2X/def2-TZVP for the QM layer and PM6 for the SE layer with the inclusion of solvation energy corrections (SMD = toluene for 3aa and 5aa; SMD = dichloromethane for 7aa). The obtained Gibbs free energies were corrected by zero-point vibrational energy (ZPVE) and temperature (268.15 K for 3aa and 7aa and 195.15 K for 5aa). The corrected Gibbs free energy of the transition state (TS1) of 3aa was calculated to be 2.06 kcal/mol lower than that of the transition state (TS2) of the enantiomer of 3aa, which corresponds to our experimental result (Fig. 5d, top). Noncovalent interactions (NCI) analysis (See the Supplementary Information for details)69 of TS2-3aa revealed unfavorable interactions between the substituents of P9 and the benzyl group of 1a. Similar transition state structures (TS1-7aa and TS2-7aa) to TS1-3aa and TS2-3aa were identified for 7aa, with TS1-7aa being 1.47 kcal/mol less energetic than TS2-7aa (Fig. 5d, bottom). The existing transition structures indicated steric interactions between the catalyst and the benzoyl group of 6a. For 2-arylindoles, the transition state structures for 5aa and the enantiomer of 5aa are shown in Fig. 5d (middle). Although the energy gap between TS1 and TS2 for 5aa (1.04 kcal/mol) was smaller than for 3aa or 7aa, it was in good agreement with our experimental result. Based on our NCI analysis, TS2-5aa exhibited unfavorable steric interactions between the catalyst and the benzoyl group of 4a. As illustrated in TS1-3aa and TS1-7aa, the methyl group at the N−1 position of 3aa and the C−3 position of 7aa are directed toward the voluminous substituent of P9, resulting in limited space between the methyl group and the catalyst. This restricted space might explain the observed low reactivity in 3pa, 3qa, and 7ha.

Biological evaluation

Given the biological application of indole scaffolds41,42,43, particularly as anticancer drugs70,71, the antiproliferative properties of 3, 5, and 7 were examined using the SRB assay (Fig. 6). To avoid any biological bias for enantiopure compounds, early tests were conducted on selected racemic mixtures. Interestingly, independent of the position of an aromatic ring, several compounds exhibit significant antiproliferative effects with an IC50 value of 10–20 μM against numerous cancer cell lines. For example, among N-arylindoles (Fig. 6a, left), (±)-3ma bearing a chloro group at the C–5 of indole exhibited promising antiproliferative effects with an IC50 of 13.4–20.8 μM. In addition, (±)-5da and (±)-7ba exhibited significant biological activities against five human cancer cell lines. This result shows that additional medicinal chemistry investigations will afford novel, promising anticancer drugs. Further biological studies of each enantiomer of 3ma were performed, as shown in Fig. 6b. The antiproliferative activity of ent-3ma, which was synthesized in the presence of ent-P9, was overall greater than that of 3ma. When both atropisomers were treated with normal human lung fibroblast cell lines to test toxicity, they exhibited approximately 2- to 3-fold lower antiproliferative activities compared to cancer cells. Although the variation in antiproliferative activity between cancer and normal cell lines was not dramatically significant, this study highlighted the significance of atroposelective synthesis of a biologically relevant scaffold and dependable atroposelective strategy72,73,74. To verify our results, the compound 3aa was exposed to Dulbecco’s Modified Eagle Medium (DMEM) medium for 72 h at 37 °C, in which any decomposition including hydrolysis was not observed. To further demonstrate the applicability of our methodology, a 1-mmol-scale reaction, and derivatizations were conducted (Fig. 6c). The reaction of 1 mmol of 1a with 2a in the presence of P9 resulted in the formation of the desired product in 74% yield and 96% ee. Subsequently, the product 3aa underwent benzylation to yield compound 8 without any loss of enantioselectivity. In addition, the oxidation of 3aa provided indole 3-carboxaldehyde compound (9), which can be further transformed into diverse compounds.

Fig. 6: Antiproliferative activities of selected atropisomeric indoles and further chemical modifications.
figure 6

a Antiproliferative activities of the racemic mixtures of selected atropisomeric indoles. aAll values are the means of at least three experiments. Measured by the Sulforhodamine-B (SRB) method. Data are represented as means ± SEM. b Antiproliferative activities of 3ma and ent-3ma. The results are representative of three independent experiments. Measured by the SRB method. Data are represented as means ± SD. c One-mmol-scale reaction and chemical derivatization of product. A549: Human lung cancer cell line; HCT116: Human colon cancer cell line; MDA-MB-231: Human breast cancer cell line; SK-Hep-1: Human liver cancer cell line; SNU-638: Human stomach cancer cell line; MRC-5: Normal human lung fibroblast cell line.

Full size image

Discussion

Chemists have made tremendous efforts to increase the availability of chiral compounds using asymmetric catalysis, which has led to the development of invaluable synthetic methods. To move on to the subsequent phase of asymmetric catalysis, we believe that one of the characteristics of a future synthetic methodology would be its high generality. It would be highly desirable that the scope covers a related but distinct type of substrates as well as a variety of substitution patterns. This objective would be achieved by developing a new strategy informed by valuable precedents. In this regard, we would like to propose a general strategy for atroposelective dynamic kinetic resolution using o-amidobiaryls, as evidenced by the asymmetric synthesis of different types of arylindoles. In the presence of chiral phosphoric acid, three types of arylindoles containing an aromatic ring in the N–1, C–2, or C–3 position exhibit high yields and high enantioselectivities (up to 99% ee). Our concept has been further demonstrated by computational studies, in which hydrogen bonding plays a crucial role in achieving high atroposelectivities. In addition, a biological evaluation of our unique scaffold reveals potent antiproliferative capabilities, allowing for the future development of an anticancer drug. We believe that our general approach might be extended to more substrate types. Furthermore, we expect that this study will serve as a stepping stone to develop a new approach for the next highest level of generalizability.

Methods

Representative procedure for the catalytic atroposelective functionalization

To an oven-dried 4 mL vial equipped with a magnetic stir bar was added substrate 1 (0.05 mmol, 1 equiv), P9 (0.005 mmol, 0.1 equiv), and 3 Å molecular sieves (40 mg). Then, the solution of 2 (0.075 mmol, 1.5 equiv) in PhMe (0.2 mL, 0.25 M) was added. The vial was sealed with a Teflon cap and further secured with Parafilm M®. The reaction mixture was stirred at –5 °C until complete consumption of 1. Then the solvent was removed in vacuo. The crude material was purified by flash column chromatography using an eluent of 5% EtOAc/CH2Cl2 to afford the desired material 3. The enantioselectivity was determined by chiral HPLC.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.