Introduction

Due to the rapid development in the last two decades1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17, allene chemistry has become an important branch in organic chemistry. Unlike other unsaturated hydrocarbons such as alkenes and alkynes, allenes are unique due to the intrinsic axial chirality. 1,3-Disubstituted allenes with synthetically useful functionalities are of particular high interest due to their common existence in natural products/drug molecules (Fig. 1)18 and their great potentials in the synthesis of natural products19,20,21,22,23,24. Thus, there is an urgent need for the development of catalytic highly enantioselective synthesis of such 1,3-disubstituted allenes.

Fig. 1
figure 1

Representative 1,3-disubstituted allenes. Natural products and a drug molecule

In view of the loosening nature of 1,3-substituted allene axial chirality spreading over three carbon atoms, the steric effect of the substrates is the major strategy used in such catalytic reactions of high ee (Fig. 2a). The highly enantioselective syntheses of such allenes with much less sterically bulky substituents are still a significant challenge and of high interest (Fig. 2b). Recently, CuBr2-catalyzed enantioselective allenylation of terminal alkynes (EATA) with aldehydes has been developed for such a purpose with very high ees and a wide scope, however, a stoichiometric amount of a sterically very bulky chiral amino alcohol, i.e., (R)- or (S)-diphenyl or dimethylprolinol, has to be applied (Fig. 2c)25,26. On the other hand, malonate is an extremely useful “sewing unit” in organic synthesis. Thus, a 1,3-disubstituted allene bearing a malonate unit with an unclosed reacting site is of high interest. Hayashi27, Imada28, and Trost29 have pioneered the Pd-catalyzed allene-asymmerization reaction of racemic non-terminal 2,3-allenol derivatives (or 2-(1,3-butadienyl) bromide) bearing a very bulky R group with rather bulky nucleophiles by applying a catalytic amount of certain chiral ligands27,28,29,30,31,32,33,34,35,36,37. However, the enantioselective reaction of substrates with R being the less sterically demanding yet very important 1°-alkyl group (Fig. 1) and unclosed parent malonate is still in its early infancy with ee of 60–70% (Fig. 2d)28,31,32. Furthermore, steric effect is also the strategy used in the catalyzed enantioselective synthesis of 4-mono-substituted 2,3-allenoates from stereo-defined enol triflates or 3-alkynoates by Frantz20, Tan38, Takemoto39, and their coworkers. This has also been observed in other development in this area40,41,42,43,44,45,46,47,48. In this paper, we wish to report a Pd-catalyzed protocol, in which the extensive prior coordination of the allene unit with palladium has to be excluded and DTBM-SEGPHOS has been identified as the best ligand, affording allene (Ra)-3 bearing the much less sterically bulky substituents in >90% ee. Such a strategy has been successfully applied to the enantioselective total synthesis of (R)-traumatic lactone (Fig. 2e).

Fig. 2
figure 2

Typical approaches to chiral 1,3-disubstituted allenes. a, b The current situation with such allene synthesis. c, d Reported strategies. e The approach developed in this study

Results

Optimization of reaction conditions

After some systematic screening on the ligands, solvents, Pd sources, and bases, we observed that the reaction of allenyl acetate 1a-OAc49,50,51 with diethyl malonate catalyzed by [Pd(π-cinnamyl)Cl)]2 and (R)-(−)-DTBM-SEGPHOS with K2CO3 as base in NMP at 30 °C afforded the 1,3-disubstituted allene (Ra)-3a-Et with 69% ee and 64% yield (Entry 1, Table 1)32 (defined as Procedure A). The reaction of allenyl carbonate 1a49,50,51 with diethyl malonate afforded (Ra)-3a-Et with a same level of ee although the yield was much higher (80%) (Entry 2, Table 1). Interestingly, the reaction following the Procedure B leading to a lower ee (60%) was observed (Entry 3, Table 1). Thus, we further adjusted the mixing sequence of the reactants and catalyst(s) and found that by using Procedure C, in which racemic allene substrate 1a was added at the end of the procedure, quite interestingly and unexpectedly, a much higher enantioselectivity of 87% was observed (Entry 4, Table 1). Running the reaction at 5 °C led to an ee of 90% and a yield of 85% for (Ra)-3aa (Entry 5, Table 1). We also evaluated the effect of the R group in malonate on the enantioselectivity (Entries 7–9, Table 1), confirming the steric effect of the parent malonate has no obvious effect.

Table 1 The optimization of reaction parameters

Substrate scope

After establishing the optimal reaction conditions, the scope of allene substrates with the much less sterically hindered yet synthetically versatile 2-non-substituted dimethyl malonate was investigated elaborately. Allenylic carbonates 1 bearing a linear 1°-alkyl (R1) group reacted with dimethyl malonate 2a to afford (Ra)-3 with excellent ee values in good to excellent yields: Entries 1–9 showed that this protocol is very general for the different length of the carbon chain of R1 ranging from Me to n-C11H23. Common and practical protecting groups for the hydroxyl group, such as Bn, α-naphthylmethyl, and TBS, are also viable in this transformation affording allenes (Ra)-3ja, 3ka, and 3la in the ee value of 91%, 91%, and 92%, respectively (Entries 10, 11 and 12, Table 2). Expectedly, 1m with a secondary R1 group (Cy) and 1p with t-Bu afforded (Ra)-3ma and (Ra)-3pa with a slightly higher enantioselectivity, indicating again that the steric effect is not the major determining factor for ee (Entries 13–14, Table 2). Due to the unclosed nature of the malonate unit, these products 3aa-3pa could be used for further sewing of useful and versatile organosegment(s) to make new chiral allenes.

Table 2 The scope of allenylic carbonates with non-substituted dimethyl malonate

Next we turned to investigate the reaction of 2,3-allenylic carbonates with more sterically hindered 2-substituted malonates (Table 3). Generally, 2-substituted malonates gave the desired (Ra)-3 with higher ee values as expected, and synthetically attractive allyl, allenyl, propargyl, 3-phenylpropargyl are all compatible. The installation of such unsaturated C–C units in chiral (Ra)-3 elaborated the practical utilities of this approach via subsequent cycloaddition reaction, radical reaction, or cascade reaction, forming other chiral chemicals52,53,54.

Table 3 The reaction of allenylic carbonates with differently 2-substituted malonates

Synthesis of racemic traumatic lactone

Such a method is applicable to the enantioselective synthesis of some naturally occurring allenes21,22,23. In order to show the potential in the syntheses of non-allene natural products, traumatic lactone was chosen as the target. In 1974, Masamune et al. isolated traumatic lactone from the roots of kidney bean (phaseolus vulgaris, beni-kintoki)55,56. The reported synthetic approaches to racemic traumatic lactone suffered from complicated starting materials, multi-steps, and harsh reaction conditions57,58,59. In addition, no enantioselective synthesis has been reported so far. Thus, we commenced to synthesize the racemic traumatic lactone (Fig. 3). Readily available racemic allenyl carbonate 1l49,50,51 reacted with malonate 2a under a set of similar reaction conditions at 25 °C with (±)-BINAP as the ligand to afford allene (±)-3la in 83% yield, which underwent alkaline hydrolysis, HOAc-mediated decarboxylation, and further alkaline hydrolysis to afford 4-allenoic acid (±)-4la in 78% yield. Then Au-catalyzed cycloisomerization of (±)-4la was applied to generate γ-butyrolactone bearing a trans C=C bond (±)-5la in 92% yield with 96:4 of E/Z selectivity24. After sequential hydrogenation of (±)-5la and Fe-TEMPO-catalyzed aerobic oxidation60,61 of the resulting hydrogenated product without purification, (±)-traumatic lactone was afforded in 79% yield.

Fig. 3
figure 3

The synthesis of (±)-traumatic lactone. Details of reagents, catalysts, solvent, and temperature are given in the reaction scheme

Enantioselective synthesis of (R)-traumatic lactone

Next, the current enantioselective protocol has been applied to the enantioselective synthesis of (R)-traumatic lactone (Fig. 4): Starting from readily available racemic allenyl carbonate 1j and malonate 2a under the optimal reaction conditions, allene (Ra)-3ja was produced with 91% ee in 84% yield, which was followed by alkaline hydrolysis and HOAc-mediated decarboxylation to afford 4,5-allenoic acid (Ra)-4ja in 91% yield. Then Au-catalyzed cycloisomerization of (Ra)-4ja generated γ-butyrolactone bearing a trans C=C bond (S,E)-5ja in 95% yield with 91% ee and 98:2 of E/Z selectivity24. After sequential hydrogenation of (S,E)-5ja and Fe-TEMPO-catalyzed aerobic oxidation60,61 of the resulting hydrogenated product without purification, (R)-traumatic lactone was afforded in 67% yield with 98% ee by one time recrystallization.

Fig. 4
figure 4

The first enantioselective synthesis of (R)-traumatic lactone. Details of reagents, catalysts, solvent, and temperature are given in the reaction scheme

Investigation of reaction mechanism

In order to rationalize the mechanism of this transformation, we conducted some control experiments (Fig. 5). A linear correlation between the ee of ligand and ee of 3aa was observed (Fig. 5a), which indicated that the active catalytic species was a monomeric palladium complex bearing a single chiral ligand62,63,64,65,66. When [Pd(π-cinnamyl)Cl)]2 with (R)-(−)-DTBM-SEGPHOS was stirred in THF for 2 h at 25 °C, after removal of THF via evaporation and recrystalization from CH2Cl2/n-hexane, (R)-(−)-DTBM-SEGPHOS·PdCl2·2CH2Cl2 was obtained as confirmed by X-ray diffraction study (Fig. 5b). However, no desired product was formed for the reaction of 1f and 2a with (R)-(−)-DTBM-SEGPHOS·PdCl2·2CH2Cl2 as the catalyst: 92% yield of 1f was recovered, indicating that (R)-(−)-DTBM-SEGPHOS·PdCl2·2CH2Cl2 is NOT the catalyst or pre-catalyst (Fig. 5c). On the other hand, the unique effect of the procedure for mixing these chemicals was re-confirmed (Fig. 5d). It has been reported that an allene unit may also function as ligand67,68. Thus, we reasoned that it is most likely that the allene unit in the stoichiometric amount of racemic 1f may compete with (R)-(−)-DTBM-SEGPHOS as ligand for extensive coordination with the palladium atom, leading to a lower ee (Procedure B). In fact, it was confirmed by the fact that a lower loading of the chiral ligand led to a lower ee (Fig. 5e). In addition, when Pd2(dba)3·CHCl3 was used instead of [Pd(π-cinnamyl)Cl)]2, the reaction gave a similar enantioselectivity albeit with a much lower yield, indicating that Pd(0) is indeed the catalytically active species and the in-situ generated Pd(0) is much more reactive69.

Fig. 5
figure 5

The studies on reaction mechanism. a The linear effect; b the complex formed from the Pd source and ligand; c the catalytic activity of this complex; d the effect of premixing; e confirming the role of the allene unit in the starting carbonate

Finally, a working model for the prediction of the axial chirality was proposed (Fig. 6). The monomeric palladium complex bearing a single chiral ligand, PdL*, would react with both (Ra)-1 and (Sa)-1 via SN2′-type oxidative addition from the back side of the C-OCO2Me bond to generate the same most favored E1-dienyl Pd, which would immediately yield the most favored delocalized intermediate α-alkynylidene-π-allyl palladium syn3-int A70 via the σ–π rearrangement involving the coordination of the terminal olefin with Pd. The front side of the terminal carbon atom in syn3-int A is blocked by L* ((R)-(−)-DTBM-SEGPHOS), thus, the malonate anion of 2 depronated by a base would attack from the back side the C–Pd bond in syn3-int A to genetate the chiral allene 3 in R configuration.

Fig. 6
figure 6

Asymmetric induction. Proposed mechanism and the model predicting the absolution configuration

Discussion

In conclusion, we have developed a non-substrate steric effect-based working strategy for the construction of allenes with non-sterically bulky substituents: due to the strong coordination ability of the allene unit with metals the exclusion of its extensive prior coordination with the in-situ generated Pd(0) should be avoid for the observed high enantioselectivity. We are actively extending this strategy to other challenging reactions constructing the allenes with less sterically bulky substituents.

Methods

The asymmetric synthesis of (R a)-3aa via Procedure C

To a dry Schlenk tube were added (R)-DTBM-SEGPHOS (70.5 mg, 0.06 mmol) and K2CO3 (276.2 mg, 2 mmol) in the glove box. Then [Pd(π-cinnamyl)Cl]2 (13.2 mg, 0.025 mmol) and 2a (263.8 mg, 2 mmol)/THF (3.5 mL) were added under nitrogen atmosphere sequentially. After being stirred at 25 °C for 30 min, the resulting mixture was stirred at 5 °C for another 10 min followed by the addition of 1a (226.2 mg, 1 mmol) and THF (1.5 mL) with stirring. After being stirred for 26 h at 5 °C, the reaction was complete as monitored by TLC. The resulting mixture was filtered through a short column of silica gel eluted with ethyl acetate (10 mL × 3). After evaporation, the residue was purified by flash column chromatography (eluent: petroleum ether (30–60 °C)/ethyl acetate = 30/1) on silica gel to afford (Ra)-3aa (217.6 mg, 77%) (eluent: petroleum ether/ethyl acetate = 30/1) as an oil: 90% ee (HPLC conditions: Chiralcel OD-H column, n-hexane/i-PrOH = 200/1, 0.5 mL/min, λ = 214 nm, tR(min) = 16.4 min, tR(major) = 17.9 min); [α]D20 = −54.2 (c = 1.08, CHCl3)); 1H NMR (300 MHz, CDCl3) δ 5.20–5.06 (m, 2H, CH=C=CH), 3.74 (s, 6H, 2 × OCH3), 3.51 (t, J = 7.4 Hz, 1H, CH), 2.61–2.54 (m, 2H, CH2), 2.00–1.90 (m, 2H, CH2), 1.43–1.19 (m, 10H, 5 × CH2), 0.88 (t, J = 6.5 Hz, 3H, CH3); 13C NMR (75 MHz, CDCl3) δ 204.0, 169.4, 169.3, 93.0, 87.3, 52.5, 51.3, 31.8, 29.12, 29.10, 29.08, 28.8, 28.0, 22.6, 14.1; IR (neat, cm−1) 2955, 2927, 2855, 1964, 1761, 1740, 1436, 1341, 1266, 1232, 1154, 1042; MS (EI, 70 eV) m/z (%) 282 (M+, 3.9), 138 (100); HRMS calcd. for C16H26O4 [M+]: 282.1831, found: 282.1830.