Stereoselective Synthesis of 3-Oxabicyclo[3.3.1]nonan-2-ones via a Domino Reaction Catalyzed by Modularly Designed Organocatalysts

Abstract

A highly stereoselective method for the synthesis of functionalized 3-oxabicyclo[3.3.1]nonan-2-one derivatives with four contiguous stereogenic centers, including one tetrasubstituted stereogenic center, was realized through an organocatalytic domino Michael-hemiacetalization-Michael reaction of (E)-3-aryl-2-nitroprop-2-enols and (E)-7-aryl-7-oxohept-5-enals followed by a PCC oxidation. Using the modularly designed organocatalysts (MDOs) self-assembled from cinchona alkaloid derivatives and amino acids in the reaction media, the title products were obtained in good yields (up to 84%), excellent diastereoselectivities (> 99:1 dr), and high enantioselectivities (up to 96% ee).

Graphical Abstract

The domino reaction is a powerful tool in organic synthesis that enables chemists to quickly assemble complex molecules with multiple stereogenic centers from relatively simple starting materials.^[1] With the advances in organocatalysis in the past two decades,^[2] numerous organocatalyzed domino reactions have also been discovered and some of them have found applications in the natural product synthesis and drug discovery^[2b,3]

(E)-2-Nitro-3-phenylprop-2-en-1-ol (1a, Scheme 1) is a very good substrate for developing domino methods since it has both an electrophilic and a nucleophilic site inside the same molecule. Indeed, several organocatalytic domino reactions have been developed based on this substrate and its derivatives.^[4] Among the reported examples, Ma and co-workers, using the Hayashi-Jørgensen catalyst 3, briefly studied the domino Michael-hemiacetalization reaction between 1a and pentanal (2) in 2011 (Scheme 1, upper equation).^[4c] The reaction produced the tetrahydropyran derivative 4 in a good yield and an excellent ee value.^[4c]

Scheme 1.

Organocatalyzed domino reactions of aldehydes and (E)-2-nitro-3-phenylprop-2-en-1-ol (1a)

Our group is interested in developing novel diastereodivergent catalytic methods^[5,6] using the modularly designed organocatalysts (M DOs),^[7] which are self-assembled in the reaction media from cinchona alkaloid derivatives and amino acids. Inspired by M a’s work,^[4c] we envisioned that a domino Michael-hemiacetalization-oxa-Michael reaction between 1a and (E)-7-oxo-7-phenylhept-5-enal (5a) catalyzed by MDOs should lead to the formation of the pyranopyran derivative 6 (Scheme 1, middle equation) and, based on our earlier results,^[5] potentially diastereodivergence could also be achieved during the product formation by using appropriate M DOs. Nonetheless, we found that the reaction did not yield the expected pyranopyran derivative. Instead, we obtained the bicyclic hemiacetal 7a as the only product, which, upon oxidation with PCC, gave the 3-oxabicyclo[3.3.1]nonan-2-one derivative 8a (Scheme 1, lower equation). Interestingly, we alsofound that the formation of 7a, most likely through a domino Michael-hemiacetalization-Michael reaction, could only be achieved by the catalysis of MDOs. Herein we wish to report the use of this unusual domino reaction between (E)-3-aryl-2-nitroprop-2-en-1-ols and (E)-7-aryl-7-oxohept-5-enals that was solely catalyzed by MDOs followed by the PCC oxidation for the highly stereoselective synthesis of 3-oxabicyclo[3.3.1]nonan-2-ones. It should be pointed out that the 3-oxabicyclo[3.3.1]nonane motif is a key structural feature of many natural products or medicinally interesting compounds and their selective synthesis have received a lot of attentions.^[8] However, assembling such bicyclic molecules with four or more contiguous stereogenic centers using a domino strategy has remained challenging despite many efforts.^[9]

At the outset, 1a and 5a was adopted as the model substrates. Several cinchona alkaloid derivatives and amino acids were adopted as the stereocontrolling modules and the reaction-center modules, respectively (Figure 1). These two modules have complementary basic and acidic functional groups that can help them self-assemble in situ in the reaction media. The most interesting results of the catalyst screening are collected in Table 1.

Figure 1.

Structure of the stereocontrolling and reaction-center modules [Ar¹ = 3,5-(CF₃)₂C₆H₃-; Ar² = 4-CF₃C₆H₄₋].

Table 1.

Catalyst screening and optimization of the reaction conditions^[a]

Entry	Modules		Time [h]	Yield [%][b]	dr[c]	ee [%][d]
1	9a	10a	24	83	99:1	61[e]
2	9a	-	24	0	---	---
3	-	10a	24	0	---	---
4[f]	-	3	24	0	---	---
5	9b	10a	20	72	99:1	71
6	9c	10a	24	50	99:1	63
7	9d	10a	20	53	99:1	7[e]
8	9e	10a	24	40	99:1	76[e]
9	9f	10a	24	18	99:1	88[e]
10	9b	10b	15	40	99:1	72
11	9b	10c	15	0	---	---
12	9b	10d	15	<5	---	---
13	9b	10e	16	72	99:1	94
14[g]	9b	10e	15	72	99:1	95
15[g,h]	9b	10e	15	72	99:1	92
16[g,i]	9b	10e	15	61	99:1	93
17[g,j]	9b	10e	15	43	99:1	93
18[g,k]	9b	10e	15	58	80:20	93
19[g,l]	9b	10e	15	40	96:4	91
20[g,m]	9b	10e	15	40	96:4	76
21[g,n]	9b	10e	15	<5	---	---

^[a]Unless otherwise specified, all reactions were carried out with 1a (0.20 mmol), 5a (0.10 mmol), and the precatalyst modules (0.010 mmol each, 10 mol%) in dry toluene (1.0 mL) at room temperature. Once the reaction was complete, the initial products were purified by column chromatography and then oxidized with PCC (3.0 equiv.) in CH₂Cl₂ at rt for 24 h.

^[b]Yield of the isolated product after flash column chromatography (overall yield after two steps of reactions).

^[c]Determined by 1H NMR analysis of the crude product.

^[d]Determined by HPLC analysis on a ChiralPak IC column.

^[e]The opposite enantiomer was obtained as the major product.

^[f]Conducted with 3 (20 mol%) and benzoic acid (60 mol%) in water (0.4 mL) at rt.

^[g]The loading of 1a was 0.12 mmol.

^[h]The reaction was conducted in 0.4 mL of toluene.

^[i]The solvent was benzene.

^[j]The solvent was xylenes.

^[k]The solvent was CHCl₃.

^[l]The solvent was THF.

^[m]The solvent was MeOH.

^[n]The reaction was conducted at 0 °C.

As the results in Table 1 show, when quinidine thiourea 9a (QDT) and L-proline (10a) were adopted as the stereocontrolling module and the reaction-center module, respectively, the reaction of 1a and 5a did not yield the expected oxa-Michael product 6; instead, the domino reaction product 7a was obtained, which gave product 8a in a good yield and an excellent dr (83%, 99:1 dr) after oxidation with PCC; nonetheless, only a moderate ee value of 61% was obtained for this major diastereomer (Table 1, entry 1). Control experiments with either 9a or 10a alone as the catalyst did not yield any product under otherwise identical conditions (entries 2 and 3). Moreover, the reaction conducted with 3 as the catalyst and benzoic acid as the cocatalyst under the conditions reported by M a and coworkers^[4c] did not yield any product, either (entry 4). On the other hand, using catalyst 3 together with DABCO as a cocatalyst only led to the formation of some unidentified intermediate that contains catalyst 3 (data not shown). These results clearly indicate that this unusual domino reaction is a unique catalytic property of the MDO. Using 10a as the reaction-center module, further screening of the other stereocontrolling modules (entries 5–9), revealed that quinidine squaramide 9b is the best stereocontrolling module in terms of both the product yield and stereoselectivities (entry 5). Using 9b as the stereocontrolling module, we then screening some additional amino acids as the reaction-center modules. Primary amino acids are not effective reaction-center modules for this reaction at all (data not shown). In contrast, secondary amino acids are more reactive (entries 10–13). As the results in Table 1 show, the MDO formed from 9b and octahydroindolecarboxylic acid 10e yielded the desired product 8a in the highest dr (99:1) and ee value (94%, entry 13). It should be pointed out that the opposite enantiomer was obtained as the major product in the cases of MDOs 9a/10a (entry 1), 9d/10a (entry 7), 9e/10a (entry 8), and 9f/10a (entry 9). On the other hand, pseudo-diastereomeric MDOs, such as 9a/10a (entry 1) and 9e/10a (entry 8), or 9b/10a (entry 5) and 9b/10b (entry 10), yielded the same enantiomer of the same diastereomeric product as the major product, and no diastereodivergence was observed. With the best MDO (9b/10e) in hand, we further optimized the reaction conditions. It was found that the loading of 1a could be further reduced to 1.2 equiv., without affecting the output and stereoselectivities of this reaction (entry 14). However, reducing the toluene from 1.0 mL to 0.4 mL resulted in a slight loss of stereoselectivities (entry 15). Common organic solvents, such as benzene (entry 16), xylenes (entry 17), chloroform (entry 18), and THF (entry 19), all yielded lower dr values and slightly lower ee values as compared to toluene (entry 14). MeOH proved to be a much poorer solvent since product 8a was obtained in a much lower ee value (76% ee, entry 20). Finally, lowering the reaction temperature to 0 °C did not improve the stereoselectivities of this reaction (entry 21).

Once the reaction conditions were optimized, the substrate scope of this reaction was established. As the results in Table 2 show, besides (E)-2-nitro-3-phenylprop-2-en-1-ol (2a), other 3-phenyl-substituted nitroalkenes are also good substrates for this reaction. The electronic nature and its position of the substituent on the phenyl ring have no effects on the diastereoselectivities (a single diastereomer was obtained in all cases) and have only a minimal influence on the product ee values (entries 2–9). The product yields were also comparable, except that slightly lower yields were obtained for the 4-methoxy-(entry 3), 2-bromo (entry 7), and 3-chloro-substituted (entry 8) substrates. Good results were also obtained for 3-(2-thiophenyl)- and 3-(1-naphthyl)-substituted nitroalkenes (entries 10 and 11), except that a slightly lower ee value (81%) was obtained for the latter. Nonetheless, an alkyl-substituted 2-nitroprop-2-en-1-ol, (E)-4-methyl-2-nitropent-2-en-1-ol, did not participate in this reaction, and no desired product was obtained (data not shown). In contrast, although the electronic nature of the substituent on the phenyl ring of (E)-7-aryl-7-oxohept-5-enals did not affect the diastereoselectivities of this reaction (a single diastereomer was obtained in all cases), they did have some influence on the product ee values: Slightly lower ee values were obtained for the electron-donating groups than the electron-withdrawing groups (entries 12–17). In addition, the yields were lower for both substrates with an electron-donating group (entry 12) and strong electron-withdrawing groups (entries 16–17). On the other hand, while 7-aryl-substituted enals are good substrates for this reaction, 7-alkyl-substituted enals, such as 4-methyl- and 7-Z-butyl-substituted (E)-7-oxohept-5-enals, do not participate in this reaction (data not shown).

Table 2.

Substrate scope the domino Michael acetalization reactions^[a]

Entry	R1	R2	Time [h][b]	8/Yield [%][c]	ee[d] [%]
1	Ph	Ph	15/24	8a/72	95
2	4-Me	Ph	72/17	8b/77	91
3	4-OMe	Ph	60/24	8c/54	91
4	4-F	Ph	48/24	8d/71	90
5	4-Cl	Ph	120/17	8e/71	90
6	4-Br	Ph	37/24	8f/84	88
7	2-Br	Ph	48/24	8g/57	94
8	3-Cl	Ph	15/17	8h/66	93
9	3-Br	Ph	15/17	8i/79	95[e]
10		Ph	38/16	8j/68	96
11	1-Nap	Ph	144/17	8k/51	81[e]
12	Ph	4-Me	15/72	81/56	80[e]
13	Ph	4-F	15/42	8m/70	94[e]
14	Ph	4-Cl	15/24	8n/73	84
15	Ph	4-Br	60/16	8o/66	87
16	Ph	4-CN	120/15	8p/55	95[e]
17	Ph	4-NO2	120/24	8q/36	94

^[a]All reactions were carried out with 1 (0.12 mmol, 1.2 equiv.), 5 (0.10 mmol) and the precatalyst modules 9b and 10e (0.010 mmol each, 10 mol %) in dry toluene (1.0 mL) at room temperature.

^[b]The second number is the reaction time for the PCC oxidation.

^[c]Yield of the isolated product after flash column chromatography (overall yield after two steps of reactions). All products were in generated with > 99:1 dras determined by 1H NMR analysis of the crude reaction products.

^[d]Unless otherwise indicated, ee values were determined by HPLC analysis on a ChiralPak IC column.

^[e]Determined by HPLC analysis on a ChiralPak ID column.

The absolute configuration of the major enantiomer obtained in this reaction was assigned on the basis of the X-ray crystallographic analysis of product 8h (Figure 2).^[10] On the basis of the X-ray data, the product was assigned the configuration of (1R,5R,6R,9S).

Figure 2.

ORTEP drawing of compound 8h (Chloroform solvate).

The domino reaction products 7 can also be converted to 3-oxabicyclo[3.3.1]nonane derivatives 11 through a dihydroxylation reaction, which also proceeds with the retention of the stereochemistry of domino products (Scheme 2).

Scheme 2.

Transformation of the domino reaction products.

On the basis of the stereochemistry outcome of this reaction, a plausible reaction mechanism is proposed to account for the formation of the major stereoisomer of this reaction (Scheme 3). The enal 5a first forms an enamine with the M DO catalyst, which then reacts with the nitroalkene 1a via a Michael addition reaction toform the intermediate 12. According to the absolute stereochemistry of the final product, intermediate 12 should have an R configuration for iminium-substituted stereogenic center and an S configuration for the phenyl-substituted stereogenic center (Scheme 3). This stereochemical outcome may be rationalized by the favored transition state for the Michael reaction between 1a and 5a (Figure 3), which is proposed based on the results of a recent computational study of our catalytic system involving the M DOs.^[11] As shown in Figure 3, the Si-Si attack of the preferred syn-(E)- enamine^[11] onto the (E) −2-nitro-3-phenylprop-2-en-1-ol yields the intermediate 12. After the regeneration of the aldehyde from the iminium hydrolysis, the intermediate has two possible pathways to cyclize to produce the final product 7a. In pathway a, the hemiacetalization of the aldehyde happens first to give the intermediate 13, which then cyclize to via an intramolecular Michael reaction to give the final product 7a. In pathway b, the intramolecular Michael addition happens first to give the intermediate 14, which then hemiacetalizes to give 7a. While the possibility of pathway b cannot be ruled out, we favor pathway a. As mentioned above, diastereodivergence was not observed in this reaction when pseudo-diastereomeric MDOs were used (Table 1, entry 1 vs. entry 8; entry 5 vs. entry 10). If pathway b is indeed involved, these results suggest a complete substrate control during the formation of the intermediate 14. However, our previous study has demonstrated that formation of compounds similar to the intermediate 14 via the carba-Michael reaction is subject to strong catalyst control and diastereodivergence can be easily achieved.^[11a] thus, we believe the lacking of diastereodivergence in the current reaction indicates that product 7a is formed through pathway a via a domino Michael-hemiacetalization-Michael reaction and the stereochemistry during the formation of 7a from 12 is totally substrate-controlled.

Scheme 3.

Plausible reaction mechanism.

Figure 3.

Proposed favored transition state that leads to the formation of the intermediate 12.

In summary, we have developed a highly stereoselective synthesis of 3-oxabicyclo[3.3.1]nonan-2-one derivatives with four contiguous stereogenic centers, including one tetrasubstituted stereogenic center, using a domino Michael-hemiacetalization-Michael reaction (E)-3-aryl-2-nitroprop-2-enols and (E)-7-aryl-7-oxo-hept-5-enals uniquely catalyzed by the MDOs followed by oxidation with PCC. Although this reaction looks like other similar reactions involving nitroalkenes and aldehydes, it is different: The reaction is not catalyzed by the Hayashi-Jørgensen catalyst and is not subject to the MDO-controlled diastereodivergent catalysis.

Experimental Section

General experimental procedure for synthesis of 3-oxabicyclo[3.3.1]nonan-2-ones via the domino Michael-hemiacetalization-Michael reaction followed by an oxidation reaction: To a vial were added sequentially the precatalyst modules 9b (6.3 mg, 0.010 mmol, 10.0 mol %) and 10e (1.69 mg, 0.010 mmol, 10.0 mol %) and dry toluene (1.0 mL). The resulting mixture was stirred at room temperature for 15 min. Compound 1a (20.2 mg, 0.10 mmol) was then added and the mixture was further stirred for 5 min. before the addition of compound 5a (21.4 mg, 0.12 mmol, 1.2 equiv.). The resulting solution was stirred at room temperature for 15 h until the reaction was complete (monitored by TLC). Then the reaction mixture was concentrated in vacuum and the residue was purified by flash column chromatography to give the hemiacetal 7a as a colorless solid (32.0 mg). A solution of the hemiacetal 7a (32.0 mg, 0.084 mmol) in CH₂Cl₂ (3.0 mL) and PCC (54.1 mg, 0.25 2 mmol, 3.0 equiv.) was stirred at room temperature for 24 h until the completion of reaction (monitored by TLC). The suspension was filtered through a short pad of silica gel and washed with ethyl acetate. Removing the solvents under vacuum afforded the crude product 8a, which was then purified by flash chromatography with 30:70 EtOAc/hexane to afford product 8a. (27.3 mg, 72%) as a white solid. m.p. 220 – 222 °C. ¹H NMR (500 MHz, CDCl₃) δ 7.93 (d, J = 7.7 Hz, 2H), 7.62 (d, J = 7.4 Hz, 1H), 7.51 (t, J = 7.7 Hz, 2H), 7.38 (dd, J = 5.1, 2.0 Hz, 3H), 7.18 (dd, J = 6.7, 2.7 Hz, 2H), 4.93 (dd, J = 12.8, 1.9 Hz, 1H), 4.72 (dd, J = 12.7, 1.9 Hz, 1H), 3.81 (s, 1H), 3.55 – 3.50 (m, 1H), 3.26 (q, J = 2.9 Hz, 1H), 3.15 (dd, J = 16.8, 10.4 Hz, 1H), 2.82 (dd, J = 16.9, 2.0 Hz, 1H), 2.33 – 2.17 (m, 3H), 1.66–1.57 (m, 1H); ¹³C NMR (125 MHz, CDCl₃) δ 195.8, 170.0, 136.3, 135.2, 133.8, 129.4, 128.9, 128.8, 128.1, 87.8, 65.2, 51.3, 43.4, 42.3, 39.0, 30.7, 26.6. V_max ( neat, cm⁻¹): 1747, 1679, 1538, 1464, 1336, 1179, 1080. HRMS (ESI): m/z calcd for C₂₂H₂₂NO₅⁺ ([M + H]⁺): 380.1492; found 380.1494. Enantiomeric excess of 8a was determined by chiral stationary phase HPLC analysis using a ChiralPak IC column (90:10 hexanes/i-PrOH at 1.0 mL/min, λ = 254 nm), major enantiomer: t_R = 35.5 min, minor enantiomer: t_R = 57.3 min.