Introduction

C-substituted morpholines and morpholinones represent important building blocks1,2 in organic synthesis and pharmacophores in medicinal chemistry, such as commercially available drugs of Finafloxacin3, Plk-2 inhibitor4 and Aprepitant5 (Fig. 1). Consequently, efficient access to these structural targets have been in great demand. Thus far, there have been a number of methods for the synthesis of C-functionalized morpholines or morpholinones6,7,8 benefiting from metal-catalyzed cyclization reactions9,10,11,12,13,14, allylic alkylation15, hydrogenations16,17 or other multi-step transformations18,19,20. On the contrary, approaches for the catalytic preparation of C3-disubstituted morpholines/morpholin-2-ones are far less developed, which is due probably to the steric hindrance effect of the incoming aza-quaternary carbon and availability of the corresponding branched starting materials21,22,23. Furthermore, asymmetric synthesis of such compounds was envisioned to be even more challenging, which must overcome the formidable challenges in both reaction reactivity and stereoselectivity. In this regard, an elegant work by Carreira and coworkers dealing with ring expansion of 3-oxetanone-derived spirocycles under In(III) catalysis is noteworthy (Fig. 2A)24. Mechanistically, the newly formed aza-quaternary carbon results from the Strecker reaction25 of oxazolidine with trimethylsilyl cyanide, and a subsequent stereoselective desymmetrizing 6-exo-tet cyclization finally affords morpholine in high dr value. To the best of our knowledge, so far, this is the only example of asymmetric synthesis of morpholine derivative with an aza-quaternary carbon center which is in situ formed rather than originates from a substrate. Thus, the highly useful and efficient approaches for the de novo construction of structurally diverse C3-disubstituted morpholines/morpholin-2-ones from readily available linear compounds are highly desirable.

Fig. 1: Selected examples of drugs containing chiral morpholine moieties.
figure 1

Chiral morpholines bearing aza-tertiary stereocenters or aza-quaternary stereocenters.

Full size image
Fig. 2: Catalytic synthesis of C3-substituted morpholines/morpholin-2-ones.
figure 2

A In(III)-catalyzed synthesis of morpholine containing an aza-quaternary center by Carreira. B Aza-benzilic ester rearrangement for the construction of aza-tertiary center of morpholinone. C Construction of morpholinone bearing aza-quaternary center by ZnCl2 catalysis (this work).

Full size image

Condensation of arylglyoxals with vicinal amino alcohols delivering 2-acyloxazolidines which further isomerize to morpholinones is well-known26,27,28,29,30. Recently, this reaction has been applied to the diastereoselective synthesis of morpholinones by Walczak group31. Very recently, we demonstrated an enantioselective aza-benzilic ester rearrangement for the synthesis of enantioenriched morpholinones by chiral Brønsted acid (Fig. 2B)32. In both cases, however, only morpholinones bearing aza-tertiary stereocenters have been successfully synthesized while all efforts aimed at the construction of C3-disubstituted morpholin-2-ones afford either oxazolidine intermediates or degradation at higher temperature31,32. To the best of our knowledge, 1,2-rearrangement for the construction of morpholinones bearing aza-quaternary stereocenters has not been achieved in spite of the significant synthetic importance of the resulting compounds33,34. In connection with our ongoing research interest in developing catalytic asymmetric cyclizative rearrangement (CACR)32,35, we herein report the successful realization of this endeavor, which delivers a wide range of C3-disubstituted morpholin-2-ones 3 from readily available linear 2,3-diketoesters 1 and vicinal amino alcohols 2 under the catalysis of ZnCl2 complex (Fig. 2C). Moreover, a highly diastereoselective version of such transformation has also been achieved by employing chiral vicinal amino alcohols as reaction partners. More importantly, rather than previously reported formal [4 + 1] oxazolidination/ring-expansive hemiacetalization/1,2-aryl or alkyl shift cascade (Fig. 2B)32,35, mechanistic investigations indicated that this transformation proceeded via a completely different reaction pathway consisting of an initial direct [4 + 2] cyclization (Int) followed by a 1,2-ester/amide shift process (Fig. 2C)36.

Results

Reaction development

With our on-going efforts to further examine its reactivity profile35,36, the multi-functionalizable vicinal tricarbonyl compounds37,38,39,40,41,42,43,44,45,46,47 were chosen as the reaction partner with aminoethanols. Our studies commenced with a mixture of ethyl 2,3-dioxo-3-phenylpropanoate and its hydrate (1a, 1.0 equiv), 2-((4-methoxyphenyl)amino)ethan-1-ol 2a (1.2 equiv) in 1,2-dichloroethane (c 0.1 M) at 80 °C under argon. Initially, screening of various strong Brønsted acids, such as trifluoroacetic acid, trifluoromethanesulfonic acid, and p-toluenesulfonic acid, failed to give the rearranged products (entries 1-3, Table 1). Additionally, even aqueous hydrochloric acid could not give the desired product. As typical bifunctional catalysts, chiral phosphoric acids (CPA)48,49 could not give the rearranged product 3a as well. Boron trifluoride-diethyl etherate (BF3 ∙ Et2O) was also ineffective (entry 4). Screening of a series of strong Lewis acids, such as Mg(OTf)2, InCl3, and Zn(II) complexes (entries 5-10), in the presence of tricarbonyl compound 1a and amino alcohol 2a at 80 °C resulted in the formation of the desired C3-disubstituted morpholin-2-one 3a, and ZnCl2 gave the best result (55% yield). Using ZnCl2 as catalyst, the reaction conditions were further optimized by varying the solvent (entries 11–13), catalyst loading (entry 14), and reaction temperature (entries 15–16). Overall, the best conditions consisted of performing the reaction in DCE (c 0.1 M) at 100 °C in the presence of ZnCl2 (0.2 equiv). Under these conditions, C3-disubstituted morpholin-2-one 3a was isolated in 61% yield (entry 15).

Table 1 Optimization of the reaction conditions.[a].
Full size table

Substrate scope exploration

With the optimized conditions in hand, this ZnCl2-catalyzed cyclizative 1,2-rearrangement was applicable to a wide range of substrates (Fig. 3). Regarding the aryl group of 3-aryl substituted 2,3-diketoesters, the presence of electron-neutral (H, Ph), electron-donating (MeO) groups and electron-withdrawing (F, CF3, CO2Et) at para and meta positions of the phenyl ring were well tolerated (3a-3j). Unfortunately, an aryl group bearing an ortho substituent failed to give the desired product which is due probably to the steric hinderance effect. The structure of compound 3a was confirmed by X-ray crystallographic analysis (see Supplementary Data 1). In the presence catalytic amount of ZnCl2, (S)-(+)-Ibuprofen derived 2,3-diketoester could give the corresponding C3-disubstituted morpholin-2-one 3g in moderate yield. Moreover, arylcoupled Aspirin could be smoothly functionalized with the present methodology (3h). 3,4-Dichlorophenyl, 2-naphthyl, and even heterocycles, such as benzo[d][1,3]dioxole, 3-thiophene and 2-thiophene, were transferred without event (3k-3o). To our delight, with aliphatic 2,3-diketoesters (1p-1s), these rearrangement reactions efficiently afforded the corresponding product 3p-3s as well. Interestingly, instead of vicinal tricarbonyl compounds, 1,2-diketone bearing CF3 group (1t)50 could also give the desired rearranged product (3t) with decreased yield. Furthermore, 2,3-diketoesters derived from different alcohols, such as methyl, isopropyl, and benzyl, were all well-accepted substrates (3u-3w). Gratefully, 2,3-diketoamides participated in this ZnCl2-catalyzed cyclizative 1,2-rearrangement to provide the corresponding rearranged products 3x and 3y in high yields. Performing the gram scale reactions of 1y (6.78 mmol) and 2a (8.14 mmol) in the presence of ZnCl2 (20 mol%) afforded 3y without obvious erosion of yield (75%). L-(-)-Menthol derived 2,3-diketoesters gave the corresponding product 3z in good yield as well. Alternatively, other N-substituted amino alcohols bearing phenyl, 2-methylphenyl, methyl and benzyl groups were also examined, affording the corresponding products 3aa-3ad without event. Notably, amino alcohol bearing electron-withdrawing protecting group on nitrogen, such as Ts and Ns, failed to afford the desired product 3. Furthermore, N-unsubstituted amino alcohols could give the corresponding rearranged products (3ae-3af) smoothly.

Fig. 3: Substrate scope of vicinal tricarbonyl compounds and amino alcohols.
figure 3

[a] Conditions: 1 (0.10 mmol), 2 (0.12 mmol), ZnCl2 (0.02 mmol), DCE (c 0.1 M), sealed tube, argon, 12 h. [b] Yield of gram scale reaction.

Full size image

Strategy extension

Asymmetric catalytic construction of multi-functionalized morpholin-2-ones bearing aza-quaternary stereocenters represents a formidable challenge. To further exploit the structural diversity of the products, we divert our attention to the diastereoselective synthesis of C3-disubstituted morpholin-2-ones by employing optically pure N-substituted 2-aminoethan-1-ols. Gratefully, without any modifications to the reaction conditions, this protocol was applicable to a wide range of chiral amino alcohols delivering structurally diverse products bearing multi-stereocenters in high stereoselectivities (Fig. 4). Various N-substituted 2-aminoethan-1-ol, such as (S)-2-aminopropan-1-ol, (R)-2-amino-3-phenylpropan-1-ol and (R)-2-amino-3-methylbutan-1-ol, underwent this rearrangement reaction smoothly affording the desired products 3ag-3ai in high diastereoselectivities. 2,3-Diketoamides 1aj and 1ak were also well tolerated leading to the formation of the products 3aj and 3ak in good yields with excellent stereoselectivities (>20/1 d.r.). Moreover, C2-phenyl substituted amino alcohol 1al afforded the product 3al in excellent d.r. value as well. N-benzyl substituted amino alcohol 1am delivered decreased stereoselectivity (6/1 d.r.). Subsequently, Other more challenging chiral amino alcohols were further examined. Interestingly, even N-PMP substituted (S)-1-aminopropan-2-ol 1an where the original stereocenter was far away from the newly formed aza-quaternary carbon could also afford the product 3an in good diastereoselectivity (4/1 d.r.). Substrates 1ao and 1ap derived from cyclic amino alcohols underwent the cyclizative 1,2-rearrangement without event, which furnished the corresponding products (3ao-3ap) bearing three stereocenters in excellent and good diastereoselectivities. The absolute configurations of the products 3ag, 3am, 3an and 3ao were determined to be (3S,5S), (3R,5R), (3S,6S), and (3 R,4aR,8aR), respectively (see Supplementary Data 2–5). Consequently, the absolute configurations for other morpholinones (3ah-3al and 3ap) were assigned accordingly (Fig. 4). Unfortunately, all efforts aimed at developing a catalytic enantioselective version of this rearrangement gave almost racemic products (see Supplementary Table S1 for details).

Fig. 4: Asymmetric synthesis of morpholin-2-ones with multi-stereocenters.[a]
figure 4

[a] Conditions: 1 (0.10 mmol), 2 (0.12 mmol), ZnCl2 (0.02 mmol), DCE (c 0.1 M), sealed tube, argon, 12 h. [b] Yield of 1.0 mmol scale.

Full size image

Mechanistic studies

A series of control experiments were performed to gain insight into the reaction mechanism. In the presence of p-toluenesulfonic acid (p-TSA), a formal [3 + 2] reaction of 1a with 2a occurred at 80 °C affording an oxazolidine compound 4a in moderate yield (Fig. 5a). Further treating intermediate 4a with catalytic amount of p-TSA (20 mol%) gave trace of the product 3a (Fig. 5a). To our surprise, however, resubmitting 4a to the standard reaction conditions also afforded only trace of the rearranged product 3a with concurrent formation of a linear compound 5 in 43% yield and some decomposition species (Fig. 5b). As 2-aminoethan-1-ol derivatives 2 are capable of serving as bidentate ligands to Zn(II)51, we further examine the reactivity of (2a)∙ZnCl2 complex. Under the catalysis of the in situ formed (2a)∙ZnCl2 complex, the reaction of the intermediate 4a gave product 3a in very low yield (10%) and a linear compound 5 was formed as a main product (Fig. 5c). Alternatively, a set of experiments were performed with 2,3-diketoester 1a and amino alcohol 2a under the standard conditions and the yields of both 3a and 4a were detected at different time. As it is seen clearly from the diagram shown in Fig. 5d, the yield of 3a increased steadily as the reaction proceeded with the concurrent formation of the low yield of 4a. These results revealed that, in this case, oxazolidine intermediate 4a resulted from formal [4 + 1] condensation of 1a and 2a is not a major reservoir for product 3a, which is different from the reported Brønsted acid catalyzed aza-benzilic ester rearrangements31,32. Notably, during the screening of the reaction conditions, another benzilic ester rearrangement product 636 was observed in the presence of catalytic amount of Zn(OTf)2 (Fig. 6a). However, ZnCl2 complex only gave a trace of compound 6 (Fig. 6b). Furthermore, resubmitting 6 to our standard reaction conditions failed to give the desired product 3a (Fig. 6c). These results indicate that an intramolecular aza-cyclization to the final product 3a via a possible in situ formed carbocation intermediate 6’ derived from dehydration of 6 could be excluded. Subsequently, we realized that, at the first step of this tandem reaction, a direct formal [4 + 2] reaction might be competitive to the [4 + 1] cyclization of 1a and 2a. All our efforts aimed at trapping the key intermediate of the rearrangement by employing various excessive nucleophilic reagents failed to give the corresponding products. To our delight, when the reaction of 1a and N-unsubstituted amino alcohol 2ae was performed at lower temperature (60 °C), a unique α-imine hemiacetal intermediate 7 was successfully isolated in good yield (Fig. 6d). Surprisingly, the imine moiety was formed from the condensation of primary amine with the carbonyl group linked to phenyl rather than with the middle one, which is also different from the previous reports31,32,35. Notably, no oxazolidine intermediate 4ae was detected throughout the whole reaction process26,27,28,29,30. Interestingly, a predictable regioisomer 8 resulted from the formal [4 + 2] reaction was actually not observed as well, which is due probably to the relatively high thermodynamic stability of the α-imine hemiacetal 7. Importantly, under the standard condition, the six-membered ring compound 7 underwent the rearrangement reaction smoothly, which delivered the corresponding product 3ae in high yield (Fig. 6e). In the absence of ZnCl2 catalyst, intermediate 7 failed to give 3ae. The reaction of 1a with another branched amino alcohol 2af gave similar results (Fig. 6f, g). The structure of the important α-imine hemiacetal 9 was further confirmed by X-ray single crystal diffraction technique (see Supplementary Data 6). These results clearly suggested that a six-membered ring intermediate resulting from a direct formal [4 + 2] cyclization should be the real reservoir for product 3. Moreover, treating compound 9 with catalytic amount of either p-TSA or (R)-TRIP-CPA at 100 °C failed to yield the product 3af, which not only underlined the great importance of ZnCl2 catalyst in triggering the key 1,2-ester shift step but also highlighted the difference in the reaction mechanism between current rearrangement and Brønsted acid-catalyzed benzilic ester-type rearrangement (Fig. 6h)31,32,35. Finally, the reaction of diphenyl-substituted 1,2-diketone 10 with the amino alcohol 2a gave neither the rearranged product nor any intermediates (Fig. 6i), which indicated that the extra ester group is of vital importance for this rearrangement. For all NMR spectra of the new compounds, see Supplementary Data 7.

Fig. 5: Control experiments.
figure 5

a–c Conversion of 4a catalyzed by p-TSA, ZnCl2 and (2a)·ZnCl2, respectively. d Kinetic curve for the formation of 3a and 4a.

Full size image
Fig. 6: Other control experiments.
figure 6

a, b Reaction of 1a and 2a catalyzed by Zn(OTf)2 and ZnCl2, respectively. c Conversion of 6 catalyzed by ZnCl2. d, e Reaction of 1a and 2ae. f–h Reaction of 1a and 2af. i Reaction of 10 and 2a catalyzed by ZnCl2.

Full size image

As the conversion of 1 and 2 to the five-membered oxazolidine intermediates 4 is extremely low, and the transformation of 4 to the products 3 is also highly difficult (see Fig. 5b–d), the compound 4 should be not the preferred intermediates under the conditions. Moreover, the isolation of the only [4 + 2] cyclization products 7 and 9 which could be smoothly transformed into the rearranged products 3ae and 3af clearly indicates the real reaction pathway as well. Therefore, on the basis of the mechanistic experiments, this rearrangement reaction is proposed to proceed via an unprecedent direct [4 + 2] cyclization/1,2-ester or amide migration sequence (Fig. 7a). Initially, the 2,3-diketoester 1a and amino alcohol 2a predominately undergo a formal [4 + 2] heteroannulation reaction to generate a cyclic α-iminium hemiacetal A. Subsequently, the key intermediate A then participate in a 1,2-ester shift reaction under the catalysis of in situ formed (2a)∙ZnCl2 complex affording the final C3-disubstituted morpholin-2-one 3a. The in situ formed a small amount of 4a is more inclined to generate compound 5 rather than the rearrangement product 3a, which is due probably to the driving force differences to these two compounds, respectively. The side product 5 might result from the carbon-carbon cleavage of intermediate B to produce a carbene species which is quenched by water and a potential hydride transfer reagent 2a52,53.

Fig. 7: Proposed reaction mechanism and stereochemical pathway.
figure 7

a The reaction of 1a and 2a was employed to demonstrate the reaction mechanism. b The stereochemical pathway for the formation of 3ah was shown.

Full size image

A plausible pathway leading to the observed stereochemical outcome is also illustrated in Fig. 7b. Because the ester moiety of the intermediate A1 is remote from benzyl group, the Re face of the iminium moiety is open to the ester group 1,2-migration, thus favorably affording 3ah. In contrast, the Si face of the iminium moiety is shielded by benzyl group in A2, thus making the 1,2-ester shift disfavored and resulting in the formation of the enantiomer ent-3ah in a minor amount. This scenario is in accordance with the configuration of the products as established by X-ray crystal structure analysis of four representative morpholines (3ag, 3am, 3an and 3ao, see Fig. 4).

In conclusion, we have developed the first examples of ZnCl2-catalyzed cyclizative 1,2-rearrangement for the asymmetric synthesis of morpholinones bearing aza-quaternary stereocenters which are far more challenging to obtain efficiently by existing methods. A series of structurally diverse C3-disubstituted morpholin-2-ones were efficiently constructed from readily available two achiral linear compounds. Importantly, on the basis of mechanistic studies, this rearrangement proceeds via an unprecedent sequence of direct formal [4 + 2] heteroannulation followed by a 1,2-ester or amide migration process, which is highly distinct from the reported reaction mechanism of aza-benzilic acid-type rearrangements.

Methods (Supplementary Methods in the SI)

General procedure for the synthesis of 3

To a mixture of 1 (0.1 mmol, 1.0 equiv), 2 (0.12 mmol, 1.2 equiv), and ZnCl2 (0.02 mmol, 0.2 equiv) was added 1,2-dichloroethane (1.0 mL) under argon. The reaction mixture was stirred at 100 °C for indicated hours. After completion of the reaction (monitored by TLC), the solvent was removed under vacuum. The residue was purified by column chromatography on silica gel, eluting with petroleum ether/ethyl acetate to afford the products 3 in moderate to high yields.