Introduction

Chiral allene moieties are present in a variety of natural products, molecular materials, and a few marketed drugs due to their unique structural features, chemical reactivity, and biological properties1,2. In addition, chiral allenes also represent a class of useful starting materials for a variety of organic transformations3. Accordingly, substantial endeavors have been made to develop methodologies for the asymmetric synthesis of chiral allenes. Classic approaches typically involve (kinetic) resolution of racemic allenes4,5 or chirality transfer from optically pure propargylic precursors6,7,8,9,10. Recent years have witnessed a tremendous development on metal-catalyzed enantioselective synthesis of chiral allenes11,12,13,14,15,16, such as nucleophilic addition to 1,3-enynes17,18,19, enantioselective functionalization of racemic allenes20,21,22, β-hydride elimination from enol triflates23, rearrangement of propargylic compounds24,25,26,27, coupling of terminal alkynes with diazo compounds28,29,30,31, and enantioselective addition of terminal alkynes to aldehydes32. In addition, several asymmetric organocatalytic protocols have also been reported, such as nucleophilic addition to activated enynes33,34, isomerization of alkynes35,36, phase-transfer-catalyzed alleno-Mannich reaction37,38, alkynylogous Mukaiyama aldol reactions39, and chiral ion-pair catalysis involving a formal propargylic carbocation from racemic propargylic alcohols40. Notwithstanding these significant advances, asymmetric synthesis of chiral allenes, particularly N-heteroaryl-substituted chiral allenes36, from readily available starting materials is particularly important and still remains a highly challenging task.

Chiral functionalized quinoline motifs are found in numerous biologically and pharmaceutically active alkaloids41. In addition, chiral quinoline is also a privileged scaffold for cinchonidine-type organocatalysts42 and ligands for transition metal catalysis43. In view of the importance of functionalized quinolines and the importance of chiral allenes, we are interested in developing a general, enantioselective protocol for the asymmetric synthesis of quinoline-substituted chiral allenes. Furthermore, quinoline-substituted allenes, which are important but synthetically challenging, are unprecedented to the best of our knowledge.

1,3-Enynes can be readily prepared via Sonogashira coupling reactions44, and are valuable potential precursors to access allenes. However, known synthetic methods to make chiral allenes from 1,3-enynes are rather limited to reactions of activated or electron-deficient 1,3-enynes17,18,19,33,34, and non-activated 1,3-enynes do not undergo these nucleophilic addition reactions due to their weak electrophilicity. Recent advances on Cu-catalyzed functionalization of alkenes indicate successful conversions of alkenes to nucleophilic alkylcopper species45,46,47,48,49,50,51,52,53,54. Furthermore, Cu-catalyzed functionalization of 1,3-enynes has also been studied (Fig. 1a)55,56. For example, Hoveyda reported a nucleophilic addition to aldehydes with nucleophiles generated from a borylcopper species and 1,3-enynes55. In 2016, Buchwald and Liu reported an asymmetric addition to ketones using nucleophiles derived from migratory insertion of 1,3-enynes to a chiral Cu-H intermediate56. Both reactions provide propargylic products selectively (Fig. 1a)55,56, although density functional theory (DFT)-calculation studies suggest a facile isomerization of initially generated propargylic copper species to allenyl copper intermediates during conversions of 1,3-enynes. Very recently, Hoveyda, Engle, and Ge groups have independently reported enantioselective copper-catalyzed hydroboration of 1,3-enynes through trapping of allenyl copper species with pinacolborane (Fig. 1b)57,58,59,60.

Fig. 1
figure 1

Copper-catalyzed enantioselective hydrofunctionalization of 1,3-enynes. a Trapping allenyl copper intermediates with carbonyls to form propargylic products. b Copper-catalyzed asymmetric hydroboration of 1,3-enynes. c This work: trapping allenyl copper intermediates with quinoline N-oxide

Full size image

We recently reported an asymmetric alkylation of quinoline N-oxides with chiral alkylcopper species60. Inspired by the suitability of quinoline N-oxides as electrophiles for alkylcopper nucleophiles, we envisioned that quinoline N-oxides could serve as proper electrophiles to react with allenyl copper species to form chiral allene products through a five-membered, rather than seven-membered, transition state (Fig. 1c). Herein, we report our studies toward the development of an enantioselective synthesis of quinolinyl-containing chiral allenes.

Results

Evaluation of reaction conditions

We began our studies on this asymmetric allenylation by identifying an effective and selective chiral copper catalyst for the reaction between but-3-en-1-yn-1-ylbenzene (2a) and quinoline N-oxide (1a). We tested copper complexes generated in situ from CuOAc and various chiral bisphosphine ligands as catalysts. The results are summarized in Fig. 2. These reactions were conducted with (EtO)2MeSiH as hydride source in cyclohexane at 5 °C for 20 h, and the allene 3a was identified as a major product for these reactions. The reaction catalyzed by the combination of CuOAc and (R)-DTBM-segphos (L1) occurred smoothly and afforded 3a in good yield (74%), but with modest enantioselectivity (46%). Similar results were obtained for the reactions when catalysts generated from CuOAc and chiral ligands L2–L5 were employed. However, the reaction conducted with CuOAc and (S,S)-Ph-BPE (L6) produced 3a in high isolated yield (90%) with excellent enantioselectivity (95%). In addition, we also studied other reaction parameters, such as solvents, temperature, and hydride sources, and found that the reactions afforded the desired product 3a either in lower yields or with lower enantioselectivities under these tested conditions (for the details, see Table S1).

Fig. 2
figure 2

Evaluation of ligands for this copper-catalyzed chiral allene synthesis. Reaction conditions: quinoline N-oxide 1a (0.400 mmol), but-3-en-1-yn-1-ylbenzene 2a (0.200 mmol), (EtO)2MeSiH (0.400 mmol), CuOAc (8.0 μmol), ligand (12.0 μmol), cyclohexane (1 mL), 5 °C, 20 h, isolated yield, and ee was determined by chiral HPLC analysis

Full size image

Scope of 1,3-enynes for Cu-catalyzed enantioselective allenylation of quinoline N-oxide

With an active chiral copper catalyst and reliable conditions identified, we studied the scope of alkyl-substituted 1,3-enynes for this Cu-catalyzed asymmetric allenylation, and the results are summarized in Fig. 3. In general, a variety of alkenyl- or alkyl-substituted 1,3-enynes (2b–2k) smoothly underwent this Cu-catalyzed enantioselective transformation to afford the corresponding allene products (3b–3k) in high isolated yields (61−89%) with good to excellent enantioselectivities (88–99%). Enynes bearing reactive groups, such as ester (3f and 3j), benzyl ether (3g), siloxy (3h), and imide (3k), are compatible with the reaction conditions. The absolute configuration of 3k was assigned as (S) by single-crystal X-ray diffraction (XRD) analysis.

Fig. 3
figure 3

Scope of aliphatic-substituted 1,3-enynes. Reaction conditions: quinoline N-oxide 1a (0.400 mmol), enyne (0.200 mmol), (EtO)2MeSiH (0.400 mmol), CuOAc (8.0 μmol), (S,S)-Ph-BPE (12.0 μmol), cyclohexane (1 mL), 5 °C, 20 h, isolated yields, and ee was determined by chiral HPLC analysis

Full size image

The scope of aryl-substituted 1,3-enynes for this reaction is summarized in Fig. 4. In general, a wide range of aryl-substituted 1,3-enynes reacted under standard condition to produce the corresponding allene products (3l–3ag) in high yields (70−95%) and with high enantioselectivities (80–96%). The data in Fig. 4 indicate that the electronic properties of the aryl groups in aryl-substituted enynes have noticeable influence on the enantioselectivity of this reaction. For example, enynes containing electron-neutral and electron-rich aryl groups reacted to afford allenes 3a and 3l–3n with high enantioselectivities (90–94%), but the reactions of enynes containing electron-deficient aryl groups (2q–2s) took place with lower enantioselectivities (80–83% for 3q–3s). The steric hindrance at the para- or meta-positions of the aryl groups in aryl-substituted enynes has little effect on the enantioselectivity of this transformation (90–94% ee for 3l–3o). However, the reaction of an enyne containing ortho-substituted aryl group (2p) occurred with significantly lower enantioselectivity (80% ee for 3p).

Fig. 4
figure 4

Scope of aromatic-substituted 1,3-enynes. Reaction conditions: quinoline N-oxide (0.400 mmol), enyne (0.200 mmol), (EtO)2MeSiH (0.400 mmol), CuOAc (8.0 μmol), (S,S)-Ph-BPE (12.0 μmol), cyclohexane (1 mL), 5 °C, 20 h, isolated yields, and ee was determined by chiral HPLC analysis. #CuOAc (16.0 μmol), (S,S)-Ph-BPE (24.0 μmol)

Full size image

This Cu-catalyzed asymmetric transformation tolerates a range of functionalities, including fluoro (3v), chloro (3w), bromo (3x), thioether (3y), tertiary amine (3z), trifluoromethyl ether (3aa), silylether (3ab), and carboxylic ester (3s, 3ac, and 3ah) moieties. 1,3-Enynes bearing nitrogen- and sulfur-containing heteroaryl substituents also reacted to produce the corresponding allenes (3ad and 3ae) in high yields and high enantioselectivity. In addition, 1,3-enynes with substitution at the olefenic terminus of the enynes ((Z)-2af and (Z)-2ag) also reacted to afford the corresponding allenes (3af and 3ag) in high isolated yields and good enantioselectivities. (Z)-2-Alken-4-ynoate (2ah), an activated 1,3-enyne, also reacted to give the allene product 3ah in high yield, albeit with a modest enantioselectivity. However, (Z)-oct-2-en-4-yne, a 1,3-enyne containing two alkyl groups at 1,4-positions, does not undergo this Cu-catalyzed asymmetric reaction.

Scope of quinoline N-oxides

Fig. 5 summarizes the scope of quinoline N-oxides for this Cu-catalyzed asymmetric allenylation transformation. In general, quinoline N-oxides with substituents at various positions reacted smoothly with the enyne 2a to yield the corresponding allenes (3ai–3ar) in high yields and high enantioselectivities. However, 8-methyl and 8-methoxyquinoline N-oxides reacted with decreased enantioselectivities (3as and 3at), which might be due to the increased steric hindrance around the N–O bond. Various reactive groups, such as chloro (3am and 3an), bromo (3ao and 3ap), siloxy (3aq), and carboxylic ester (3ar), at different positions of quinoline N-oxides were tolerated, and the allene products were isolated in high yields (61–97%) with high to excellent enantioselectivities (84–97% ee). However, the allenylation reactions of N-oxides of pyridines and isoquinolines do not occur under identified conditions.

Fig. 5
figure 5

Scope of quinoline N-oxides. Reaction conditions: quinoline N-oxide 1a (0.400 mmol), enyne (0.200 mmol), (EtO)2MeSiH (0.400 mmol), CuOAc (8.0 μmol), (S,S)-Ph-BPE (12.0 μmol), cyclohexane (1 mL), 5 °C, 20 h, isolated yields, and ee was determined by chiral HPLC analysis

Full size image

Synthetic potential

To show the utility of this asymmetric protocol, the reaction of quinoline N-oxide 1a with enyne 2a was conducted on gram scale with 2 mol% copper catalyst at 5 °C. This reaction proceeded to full conversion of 2a in 20 h and afforded allene 3a in 89% isolated yield and with 93% ee (Fig. 6a). The enantioselectivity of this gram-scale reaction is comparable to that of a 0.200 mmol scale reaction (Fig. 2). In addition, we also showed that the quinoline-substituted allene products 3a and 3d underwent NIS-initiated cyclization to generate the corresponding 2-iodo-pyrrolo[1,2-a]quinolones in good yields (Fig. 6b, NIS = N-iodosuccinimide)61.

Fig. 6
figure 6

Synthetic applications. a Gram-scale reaction. b NIS-promoted cyclization of quinoline-substituted allenes. c–e Allenylation reactions with chiral allenes derived from vitamin E and cholesterol. Yields are reported for isolated materials after column chromatography on silica gel

Full size image

This copper-catalyzed asymmetric allenylation was utilized for the modification of complex molecules. For example, the allenylation of quinoline N-oxide 1a with enantiometrically pure 1,3-enynes 5 and 7 (5 derived from vitamin E and 7 derived from cholesterol, respectively) occurred smoothly in the presence of 4 mol% CuOAc/(S,S)-Ph-BPE, affording chiral allenes 6 and 8 in high isolated yields with excellent diastereoselectivity (dr > 20:1), respectively (Fig. 6c, d). The corresponding reaction with enyne 7 conducted with CuOAc/(R,R)-Ph-BPE produced the opposite diastereomers 8′ in comparable yield and with similarly high diastereoselectivity. The results of this experiment indicate a mechanism involving catalyst control rather than substrate control when enantiometrically pure 1,3-enynes are used. The absolute configuration of compounds 6, 8, and 8′ was assigned by analogy to that of compound 3k.

Mechanistic considerations

To gain some mechanistic insights into this Cu-catalyzed asymmetric allenylation of quinoline N-oxides, we conducted a series of deuterium-labeling experiments (Fig. 7). The reaction of quinoline-2-D N-oxide 1a-d1 with 1,3-enyne 2a under standard conditions afforded 3a in high isolated yield after purification with flash chromatography on silica gel, but no deuterium was incorporated into the product (Fig. 7a). When conducted with Ph2SiD2, the reaction of quinoline N-oxide 1a with 1,3-enyne 2a produced 3a-d1 in high isolated yield after purification with flash chromatography on silica gel, and the deuterium was located in the methyl group of 3a (Fig. 7b).

Fig. 7
figure 7

Mechanistic studies. a–d Deuterium-labeling experiments. e Proposed catalytic pathway for this Cu-catalyzed asymmetric allenylation reaction

Full size image

To detect the potential intermediates for this Cu-catalyzed asymmetric allenylation, we monitored by 1H nuclear magnetic resonance (NMR) analysis the reaction of quinoline N-oxide 1a and enyne 2a under standard conditions, and found that this reaction afforded I-3a with a modest diastereoselectivity (dr = 3.6:1) in high yield prior to the isolation of final product by flash choromatography on silica gel (Fig. 7c). Correspondingly, the reaction of quinoline-2-D N-oxide 1a-d1 with 1,3-enyne 2a produced the intermediate I-3a-d1 in 95% yield with a diastereoselectivity of 3.6:1 (Fig. 7d). Upon workup by column chromatography on silica gel, both intermediates I-3a and I-3a-d1 underwent re-aromatization to yield 2-allenyl quinoline 3a in high yields.

Based on the precedent of reactions of quinoline N-oxides with metal-alkyl reagents62,63, the Cu-catalyzed hydrofunctionalization of 1,3-enynes57,58,59, and the results of the above-mentioned mechanistic experiments, we proposed a plausible catalytic pathway for this Cu-catalyzed asymmetric allenylation of quinoline N-oxides (Fig. 7e). Activation of CuOAc with (EtO)2MeSiH in the presence of (S,S)-Ph-BPE (L*) produces a chiral copper hydride species (L*)Cu-H (A). Enantioselective migratory insertion of 1,3-enyne into (L*)Cu-H forms a propargylic copper intermediate B, which then undergoes a stereospecific isomerization to an allenyl copper species C. This allenyl copper intermediate reacts with quinoline N-oxide to form a copper species D, which reacts with (EtO)2MeSiH to afford a siloxane intermediate E and to regenerate the catalytically active Cu-H species A. The siloxane intermediate E then aromatizes to form the 2-allenyl quinoline product.

DFT calculations

With DFT calculations, we explored the mechanism and origins of chemoselectivity based on the reaction of quinoline N-oxide 1a and 1,3-enyne 2a catalyzed by the combination of CuOAc and (S,S)-Ph-BPE. The computed free energy diagram of the most favorable pathway is shown in Fig. 8a (see Supplementary Figure 18 for the detailed calculations on other pathways). From the Cu-H species int1, the hydrocupration of 2a occurs via TS2 and generates the propargyl copper intermediate int3 enantioselectively. The origins of this enantioselective hydrocupration step have been discussed, by Buchwald and Liu, for a Cu-catalyzed asymmetric addition of olefin-derived nucleophiles to ketones51. This hydrocupration step determines the enantioselectivity of this Cu-catalyzed allenylation reaction. From int3, the stereospecific 1,3-isomerization via TS4 to form int5 is facile. Thus, the propargyl copper species int3 and the corresponding allenyl copper species int5 are in a fast equilibrium. The final C–C bond formation proceeds through the five-membered ring transition state TS7 leading to the intermediate int8. Further exchange with (EtO)2MeSiH affords int9 and regenerates the Cu-H species int1 for the next catalytic cycle. Based on the free energy changes of the whole catalytic cycle, the allenyl copper intermediate int5 is the on-cycle resting state, and the rate-determining step is the regeneration of active catalyst int1 via TS9, which is consistent with previous related experimental mechanistic results64.

Fig. 8
figure 8

DFT studies. a DFT-computed free energy changes of the most favorable pathway for the Cu-catalyzed enantioselective allenylation of quinoline N-oxide 1a with 1,3-enyne 2a. b A competition experiment between sterically differentiated quinoline N-oxides

Full size image

In addition, we also discovered an epimerization pathway from int5 to intS4 via TS-S5 (depicted in Fig. 8a), which could be responsible for the varied enantioselectivities obtained for the reactions with different quinoline N-oxides (Fig. 5). For example, 8-methyl quinoline N-oxide reacted with a significantly lower enantioselectivity (61% ee for 3at in Fig. 5) than quinoline N-oxide (95% ee for 3a in Fig. 2). We attributed this decreased enantioselectivity to a slower rate for the reaction of 8-methyl quinoline N-oxide, caused by the increased steric hindrance around the N–O bond. To further prove the difference in reactivity, we conducted a competition experiment of 1,3-enyne 2a between quinoline N-oxide and 8-methyl quinoline N-oxide in a single reaction vessel (Fig. 8b). Indeed, the allene product (3av) from the reaction of 8-methyl quinoline N-oxide was produced in a much smaller quantity than the product (3a) from the reaction of quinoline N-oxide.

The C–C bond formation step determines the chemoselectivity between allene and alkyne products. We carefully examined the possible transition states for the C–C bond formation, and the key competing ones are shown in Fig. 9. The propargyl copper intermediate int3 is 4.0 kcal/mol unfavorable as compared to the allenyl copper intermediate int5. This thermodynamic difference also reflects in the corresponding transition states, and TS12 and TS13 (from int3) are less favorable than TS7 and TS11 (from int5). Comparing TS7 and TS11, the calculations suggest a strong selectivity toward the formation of allene product via TS7, which is consistent with the experimental selectivity (Fig. 2). TS7 is a five-membered ring transition state, while the competing TS11 requires a seven-membered metallacycle, which is unfavorable due to the ring strain. The important role of metallacycle in determining the chemoselectivity provides insights for the reaction design with allenyl and propargyl copper species.

Fig. 9
figure 9

Optimized structures and relative free energies of competing transition states for C–C bond formation

Full size image

Discussion

In summary, we report an efficient and enantioselective protocol for the asymmetric synthesis of trisubstituted allenes through a Cu-catalyzed allenylation of quinoline N-oxides with 1,3-enynes. A wide range of aryl- and alkyl-substituted 1,3-enynes reacted with quinoline N-oxides to afford the corresponding allenes in high isolated yields with good to excellent enantioselectivity. This asymmetric transformation could be readily conducted on a gram scale with a reduced catalyst loading (2 mol%), and thus this protocol presents a practical route to prepare enantiomerically enriched quinoline-containing trisubstituted allenes from readily accessible quinoline N-oxides and 1,3-enynes. DFT calculations suggest that this Cu-catalyzed allenylation occurs through a nucleophilic attack of quinoline N-oxide with an allenyl copper intermediate via a five-membered transition state, thereby providing a rationale to explain the obtained chemoselectivity toward an allene rather than a propargylic product. These findings will enable future development of new reactions with an allenyl or a propargyl copper intermediate, and these studies will be the subject of future work.

Methods

General procedure for the synthesis of chiral allenes

CuOAc (4 mol%, 0.9 mg) and (S,S)-Ph-BPE (6 mol%, 6.1 mg) were dissolved in cyclohexane (1 mL) in a screw-capped vial under argon atmosphere. The resulting mixture was stirred for 5 min, to which was added (EtO)SiHMe (0.400 mmol, 65 μL), 1,3-enynes (0.200 mmol), and quinoline N-oxide (0.400 mmol) successively. The vial was stirred at 5 °C for 20 h. The mixture was concentrated under vacuum and the crude product was purified by flash chromatography on silica with a mixture of ethyl acetate/hexane (1:100 to 1:50). The ee values were determined by high-performance liquid chromatography.

(Note: The allene products are unstable and prone to racemization in acidic solvent such as CHCl3. The ee values and optical rotation should be determined immediately after isolation.)

Mechanistic studies

Spectra pertaining to deuterium labeling, competition, and other mechanistic studies are available in Supplementary Figures 1-16.

Synthesis and characterization

Full synthetic procedures are available in the Supplementary Methods. The XRD structure of 3k is available in Supplementary Figure 17 and Supplementary Data 1-2. NMR spectra of purified organic compounds are available in Supplementary Figures 19-128.

Screening conditions

Further details of reaction conditions screened are available in Supplementary Table 1.

Computational study

Full procedures and results are available in the Supplementary Methods, Supplementary Figure 18, and Supplementary Table 2. Cartesian coordinates of calculated intermediates and transition states are available in Supplementary Table 3.