Proof-of-principle study and optimization of the reaction conditions

The success of this strategy relies on exploiting a cycloisomerization of chiral δ-amino α,β-unsaturated ketones 1 to obtain 1,2-dihydropyridines. The condensation of amines with ketones usually takes place under acid catalysis²⁹. However, with a bulky acid-sensitive N-Boc protecting group, the cycloisomerization of 1 deemed to be not easy. There appears to be only one earlier report on a similar transformation; Donohoe et al. utilized a stoichiometric amount of trifluoroacetic acid (TFA) to promote the cycloisomerization of racemic N-tosyl δ-amino enones without a γ-substituent, at 80 °C³⁰. In our study, the use of 100 mol% TFA to mediate the cycloisomerization of 1a at 80 or 40 °C for 10 h resulted in ca. 7% NMR yield of target 2a, with 63% or 37% conversion of 1a and 33% or 14% NMR yield of pyridine by-product 3a (see Supplementary Table 1 for details) due to the deprotection of the N-Boc group (Fig. 2a).

Fig. 2

Proof-of-principle study.

a Optimization of conditions for the cycloisomerization of 1a. b Control experiments and possible mechanism. c Gram-scale asymmetric Mannich/Wittig/cycloisomerization sequence. Boc: tert-butyloxycarbonyl. EtOAc: ethyl acetate.

Further screening of typical Brønsted acids revealed that only HCl could promote the reaction to give the desired 2a in 24% isolated yield, along with 3a in 46% NMR yield. The generation of 3a should be achieved due to N-Boc deprotection, but almost no reaction occurred when using 20 mol% HCl. To suppress the side deprotection, it was therefore necessary to identify a milder acid catalyst. Despite several efforts, including the use of more than 20 metal salts, no mediation of this cycloisomerization was achieved (see Supplementary Tables 1 and 2 for details of optimization). Then, inspired by the advances in Lewis base activation of SiCl₄ for carbonyl transformations³¹, we tried the use of SiCl₄ to mediate the reaction, since our designed tandem sequence will stoichiometrically produce Ph₃PO. Pleasingly, the merger of SiCl₄ and Ph₃PO (each 100 mol%) mediated the reaction to afford 2a in 69% yield, which was further improved to 83% if reducing the SiCl₄ to 30 mol%. The presence of Ph₃PO benefitted the yield, since the use of 100 mol% or 30 mol% SiCl₄ alone resulted in lower yields of 2a. The merger of HSiCl₃ with Ph₃PO also led to the formation of chiral 2a in 74% yield, and no conjugate reduction of enone 1a occurred³².

The high yield achieved from the combination of SiCl₄ and Ph₃PO in this cycloisomerization prompted us to then study the reaction mechanism. Based on available knowledge, two possible roles of SiCl₄ could be rationalized, despite it waits for further studies when and how the alkene isomerization proceeds (Fig. 2b). One is that it serves as a Lewis acid catalyst, activated by Lewis base Ph₃PO, to promote the reaction (path A), and the other is that it acts as a hidden Brønsted acid^33–35, releasing HCl as the true catalyst via hydrolysis (path B), since HCl could mediate the reaction as well (Fig. 2b). However, it was found that the addition of either 0.2 or 1.0 equiv of noncoordinating base 2,6-di-tert-butylpyridine terminated the reaction, and ²⁹Si NMR analysis showed almost no interaction between it and SiCl₄ (see Supplementary Fig. 1 for details). These findings implied that the HCl generated via the hydrolysis of SiCl₄ was the true catalyst. Furthermore, it was found that the addition of 10, 30, 50, and 100 mol% H₂O to the reaction mediated by 30 mol% SiCl₄ and 1.0 equiv Ph₃PO led to a gradually decreased yield of 2a. Similarly, using 100 mol% HCl as the catalyst led to only 37% yield in the presence of Ph₃PO. This implied that the more HCl present at the initial stage of the reaction, the lower the yield of 2a due to more side deprotection. These findings agreed with the concept of hidden Brønsted acid catalysis, as the gradual generation of HCl from SiCl₄ inhibited side deprotection to improve the yield. In this cycloisomerization, the hydrolysis of SiCl₄ by the trace H₂O in the reaction system produced HCl to initiate the reaction, and the subsequent dehydration generated H₂O that was internally used to hydrolyze SiCl₄ (Fig. 2b). Such reuse of the by-product H₂O for the gradual generation of HCl as the reaction proceeded meant that the concentration of HCl was kept at a low level, which effectively suppressed side reactions. The role of Ph₃PO in improving the yield was possibly due to its acid–base interaction with HCl, which modulated the acidity of HCl, as confirmed by ³¹P NMR analysis (see Supplementary Figs. 2 and 3 for details).

Having established the conditions for the key cycloisomerization step, we next integrated it with the proline-catalyzed Mannich/Wittig reaction into a tandem sequence. The procedure finally proved to be convenient. The Mannich reaction of aldimine 4a and propanal 5a mediated by 20 mol% L-proline was performed in MeCN. On completion of this reaction, the crude solid adduct 7a was easily collected and subjected to the reaction with phosphorane 6a in EtOAc at 80 °C. Once the synthesis of N-Boc δ-amino enone 1a was completed, 30 mol% SiCl₄ was directly added to facilitate the cycloisomerization, at 40 °C. Remarkably, this sequence could be run at a 20 mmol scale, giving 1,2-dihydropyridine 2a (4.05 g) in 71% overall yield and >99% ee (Fig. 2c).

Notably, this sequence constitutes a rare example of an asymmetric tandem reaction that internally recycles waste to facilitate the downstream step^36–47: by-product water from the imine formation step was reused to produce the true catalyst HCl and waste Ph₃PO from the Wittig step was recycled as a modulator. In particular, the gradual reuse of byproduct with the proceeding of the reaction, to generate the true catalyst in situ at a low concentration to suppress side reactions, is unknown. This suggests such sustainable sequential reactions are worthwhile to explore.

Scope of asymmetric Mannich/Wittig/cycloisomerization sequence

Next, the generality of the synthesis of 2,3,6-trisubstituted chiral 1,2-dihydropyridines with respect to differently substituted imines, aldehydes, and phosphoranes was determined (Fig. 3a). A wide range of aryl N-Boc aldimines were compatible with the sequence with propanal 5a and acetone-derived ylide 6a. Regardless of the nature and position of the substituent on the phenyl ring, the desired adducts 2b–i were obtained in 43–63% yields with 96–>99% ee. 2-Naphthyl-, 2-thienyl-, and 2-furanyl-substituted N-Boc aldimines also readily afforded the desired products 2j–l in 56–66% yields with 98% ee. The scope of aldehydes 5 was satisfactory; the targets 2m–y were obtained in 48–70% yields with 92–>99% ee. This enabled the facile incorporation of diverse substituents on the C3 position of chiral 1,2-dihydropyridines (including aryl, alkyl, allyl, and (methylthio)methyl groups)—offering the hope of further modification. Moreover, α-imino ester was a viable substrate⁴⁸; it afforded 2-ethylcarboxylate-substituted 1,2-dihydropyridines 8a–c in 33–65% yields with 89–92% ee. Ethyl-, propyl-, and phenyl-substituted phosphoranes 6 were also tolerated under optimized conditions, giving the 2,3,6-trisubstituted dihydropyridines 8d–f in moderate yields with 93–99% ee. This permitted a variation of the C6 substituent of the products.

Fig. 3

Scope of asymmetric Mannich/Wittig/cycloisomerization sequence.

Reaction conditions. Mannich step: imines 4 (0.5 mmol), aldehydes 5 (1.0 mmol), L-proline (20 mol%), CH₃CN (10 mL), 0 °C, 20 h. Wittig step: ylides 6 (1.0 equiv based on the crude 7), EtOAc (2 mL), 80 °C, 15 h. Cycloisomerization step: SiCl₄ (30 mol%), EtOAc (2 mL), 40 °C, 2–10 h. Unless otherwise noted, overall yields for 3 steps are given. Ee values were determined by chiral HPLC analysis. *These reactions begin from the corresponding isolated chiral β-amino aldehydes 7 and the yields for two steps are given. †Using 1.0 equiv of 5. ‡Using SiCl₄ (45 mol%) instead of SiCl₄ (30 mol%). For details, see the Supplementary Information. a Synthesis of 2,3,6-trisubstituted 1,2-dihydropyridines. b Synthesis of 2,3- and 2,6-disubstituted 1,2-dihydropyridines. c Synthesis of 2-substituted 1,2-dihydropyridines. d Typical examples with isatin-derived N-Boc ketimines.

The modular synthesis of enantioenriched 2,3- and 2,6-disubstituted 1,2-dihydropyridines was also achieved by using acetaldehyde-derived ylide 6 or/and acetaldehyde, as demonstrated by the preparation of 9a–l with 86–>99% ee (Fig. 3b). As anticipated, the sequence in which use was made of N-Boc aldimines, acetaldehyde-derived ylide 6, and acetaldehyde 5 enabled the smooth preparation of N-Boc 2-aryl-substituted 1,2-dihydropyridines 10a,b in 32–42% yields with 97–99% ee (Fig. 3c). Remarkably, the isatin-derived N-Boc ketimines have emerged as useful substrates, as well as in the current sequence⁴⁹, affording the valuable chiral dihydropyridine-based spirooxindoles 11a–e featuring a tetrasubstituted carbon stereocenter with 85–88% ee, albeit in modest yields (Fig. 3d).

Synthetic utility

This sequence was further extended for the synthesis of polysubstituted pyridines, because the addition of MeOH to quench the remaining SiCl₄ and its hydrolyzed derivatives would release HCl that could be reused for N-Boc deprotection⁵⁰. The resulting unprotected 1,2-dihydropyridines could then be easily oxidized to afford pyridines⁵¹. Interestingly, the tunable synthesis of both pyridines and 3-hydroxyl pyridines was achieved by varying the quenching procedures (See the Supplementary Information for details). More importantly, the synthetic value of this protocol was highlighted by the diverse product transformations. First, the thus obtained structurally diverse enantioenriched 1,2- dihydropyridines could be used for the facile synthesis of piperidines with one, two, or three stereocenters, via hydrogenation, using Pd/C or PtO₂·H₂O as the catalyst (Fig. 4a). For example, N-Boc 2,3,6-cis-trisubstituted piperidines 12a–l were obtained in 75–99% yields, and with 5:1 to >20:1 dr and 87–>99% ee. The relative and absolute configuration of product 12j was assigned based on its hydrochloride 13. These piperidines had substantial structural diversity, because the C2 substituent could be a phenyl, 2-furanyl, or ester group, whereas the C3 substituent could be an aryl or aliphatic group. It should be mentioned that methods to access chiral C3 substituted azacycles are very limited⁵². Furthermore, optically active cis-polysubstituted piperidines are generally difficult to access in excellent ee values^53–59. Given the importance of piperidines in drug discovery⁴, these chiral piperidines should be very useful/valuable in medicinal chemistry research.

Fig. 4

Synthetic utility.

a Diverse synthesis of cis-polysubstituted chiral piperidines 12. b Synthesis of chiral tetrahydropyridine 14. c Diels-Alder reaction to bridged rings 15. †Using Pd/C instead of PtO₂·H₂O.

The 2,6- and 2,3-cis-disubstituted N-Boc piperidines could also be readily accessed via the Pt₂O-catalyzed hydrogenation of the corresponding chiral 1,2-dihydropyridines, as exemplified by the facile access to piperidines 12m–q in 70–95% yields, and with 12:1–>20:1 dr and 95–>99% ee values. Furthermore, starting from 2-phenyl dihydropyridine 10a, the corresponding piperidine 12r was obtained in 92% yield with 99% ee; it is a key intermediate for NK₁ receptor antagonist synthesis^60,61. Notably, chiral 1,2-dihydropyridine 2a could be selectively reduced to enantioenriched tetrahydropyridine 14 in 74% yield with >20:1 dr and 98% ee value (Fig. 4b). In addition, upon treatment with N-methylmaleimide, chiral 1,2-dihydropyridines 9g and 10a could smoothly deliver the corresponding bridged rings 15a and 15b in high yields with excellent dr and ee values (Fig. 4c). The structure and configuration of 15b was confirmed by its X-ray analysis (see Supplementary Table 17 for details).