Introduction

Asymmetric dihydroxylation of alkenes is one of the cornerstone reactions in organic synthesis, as it provides a direct entry to optically active vicinal diols, which are a subunit in a large number of naturally occurring compounds and also important building blocks in many syntheses1,2. The classic Os-catalyzed asymmetric syn-dihydroxylation, known as Sharpless dihydroxylation, demonstrates high efficiency and enantiocontrol for a broad range of substrates, and thus finds widespread applications in the total synthesis of natural products1,3,4,5,6,7,8,9,10. However, the toxicity, volatility, and high cost of OsO4 urge organic chemists to establish alternative catalytic systems for asymmetric dihydroxylation utilizing inexpensive and less toxic catalysts. In the past two decades, a number of Os-free asymmetric syn-dihydroxylations have been developed11, including the examples using chiral Mn- 12,13 or Fe- 14,15 complexes, bimetallic nanoclusters16 and organic phase transfers as catalysts17,18. In the field of anti-dihydroxylation, highly enantioselective variants are scarce. Jørgensen et al. reported a formal asymmetric anti-dihydroxylation of α,β-unsaturated aldehydes by merging amine-catalyzed enantioselective epoxidation and strong base-mediated methanolysis in a one-pot procedure (Fig. 1a)19. Moreover, Li and his co-workers applied successfully the combination of monooxygenase and epoxide hydrolase in the enzyme-catalyzed asymmetric anti-dihydroxylations (Fig. 1b)20,21,22,23. On the other side, molybdenum catalysis finds wide applications in oxidation reactions24,25. Recently, our group discovered that MoO2(acac)2 is able to catalyze direct anti-dihydroxylation of allylic alcohols with hydrogen peroxide as oxidant, providing diverse racemic 1,2,3-triols as products26,27. The challenge of realizing the highly enantioselective version of this reaction lies in not only the enantiocontrol of the initial epoxidation but also the regiocontrol of the following ring opening reaction. The lack of control in either of these two elementary steps will result in low level of asymmetric induction of the final products. Furthermore, the ligand-mismatch effect in the potential kinetic resolution in the ring opening process can also erode the high enantioselectivity obtained for the epoxides intermediates. Herein, we report a chiral molydenum-bishydroxamic acid-catalyzed asymmetric anti-dihydroxylation of allylic alcohols, providing an efficient entry to a variety of 1,2,3-triols in high enantioselectivities (Fig. 1c). Notably, the 1,2,3-triol moeity is contained as a structural motif in numerous natural products and synthetic biologically active compounds (Fig. 1d).

Fig. 1
figure 1

Anti-dihydroxylation of alkenes. a Amine-catalyzed asymmetric formal anti-dihydroxylation. b Enzyme-catalyzed asymmetric anti-dihydroxylation. c This work: molybdenum-bishydroxamic acid-catalyzed asymmetric anti-dihydroxylation. d Examples of natural products and synthetic biologically active compounds containg a 1,2,3-triol unit

Results

Optimization

For optimization of the reaction conditions, we used commercially available 3-methylbut-2-en-1-ol (1a) as the standard substrate (Fig. 2). Initially, a series of chiral bishydroxamic acids (BHA) L1-6 bearing trans-cyclohexane diamine scaffold were investigated as ligands in the reaction using MoO2(acac)2 as catalyst in EtOAc at 40 °C (Table 1, entries 1–6). In all these cases mentioned above, the reactions afforded a mixture of the desired product 2a and a hydroperoxide 2a-1, and after reductive work-up with diphenyl sulfide the triol 2a was obtained in good to excellent yields. However, in the most cases the enantiocontrol was low (entries 1–5). Only in the case of the ligand L6, the reaction furnished the product in an excellent enantiomeric excess (entry 6). Notably, opposite enantioselectivities were achieved even for the reactions using the ligands with the same absolute configuration. Next, a brief solvent and Mo-precatalysts screenings were undertaken (entries 7–19), providing no better outcome in terms of enantioselectivity. Reducing the catalyst loading to 5 mol% resulted in a diminished yield (entry 20). Finally, excellent result with respect to both efficiency and selectivity was obtained, when the reaction was conducted in lower concentration with extended reaction time (entry 21).

Fig. 2
figure 2

Model system used for optimization of the reaction conditions. Full conditions are reported in Table 1

Table 1 Ligands, solvents and Mo-precatalyst screening for the Mo-catalyzed asymmetric anti-dihydroxylation of 3-methylbut-2-en-1-ola

Substrate scope

With the optimal reaction conditions in hand, we started to evaluate the substrate scope of this Mo-catalyzed asymmetric anti-dihydroxylation of allylic alcohols (Fig. 3). First, various symmetric β,β-disubstituted allylic alcohols 1a-g were employed as substrates, affording the products 2a-g in high to excellent yields and enantioselectivities. When the substituents on the β-position are not identical, both E- and Z-alkenes turned out to be suitable precursors for this Mo-catalyzed reaction, furnishing the products 2h-u with high efficiency and excellent enantiocontrol. It was observed that a series of functional groups were tolerated including carbamate (2h), halide (2k, 2l, and 2s), imide (2m), silyl ether (2n and 2t), and ester (2o). Moreover, a complete proximal selective dihydroxylation was achieved, when the allylic alcohols bearing at least one distal olefinic unit (2o-r, 2u and 2v) were used as substrates. Remarkably, the oxidation of the distal C-C double bond of these substrates is favored in the case of Sharpless dihydroxylation6, indicating that our method is complementary to the Os-catalyzed syn-dihydroxylation concerning both diastereo- and regioselectivity. Furthermore, 1,1,2-trisubstituted and disubstituted alkenes turned out to be less reactive and thus required higher catalyst loading (10 mol%). Under the modified conditions, the reactions proceeded smoothly, affording the products 2w-ae in good to high yields. Good to excellent asymmetric induction was achieved for the erythro-triols 2y-ac starting from disubstituted E-olefins. Notably, only low to moderate enantioselectivities (<74% ee) can be obtained for the synthesis of erythro-triols using asymmetric Os-catalyzed syn-dihydroxylation of Z-allylic alcohols5. One limitation of our method was observed in the case of 1,1,2-trisubstituted and disubstituted Z-olefins, as relatively low enantiomeric excesses were obtained. Of note is that the reactions involving the alkenes 1i-ae delivered all the corresponding products in high diastereoselectivities (dr > 95:5, determined by 1H-spectroscopy). Moreover, a 5-mmol-scale reaction using the alcohol 1o was conducted, providing the product 2o in a similar yield and enantioselectivity. In this case we discovered that the Mo-BHA-catalyst decomposed completely on silica gel, releasing the chiral BHA-ligand, which was isolated in an excellent recovered yield (95%) through column chromatography and determined to be analytically pure. The reaction using the recycled ligand and MoO2(acac)2 afforded a similar result with no decrease of the enantiomeric excess.

Fig. 3
figure 3

Evaluation of the substrate scope of the Mo-BHA-catalyzed asymmetric anti-dihydroxylation of allylic alcohols. i–iii (i) Unless otherwise specified, reactions were performed on a 0.2 mmol scale of the allylic alcohols 1 using 2.5 equiv 35% aqueous H2O2, 5 mol % MoO2(acac)2 and 6 mol% L6 in 2.0 mL EtOAc at 40 °C. ii Yields of the isolated products after column chromatography. iii The enantiomeric excesses were determined by HPLC-analysis on chiral stationary phase. iv Reaction time: 4 h. v Reaction temperature: 30 °C. vi Reaction was performed on a 5 mmol scale of the allylic alcohol 1o. vii Reaction performed with the recycled BHA L6. viii Reaction time: 1.5 h. ix Catalyst loading: 10 mol% MoO2(acac)2 and 12 mol% L6. x Reaction performed in 1,4-dioxane

Some derivatizations based on the conversion of the alcohol moiety were carried out (Fig. 4). First, selective tosylation of the primary alcohol group of the triol 2r followed by SN2 nucleophilic substitution by benzyl amine provided an amino alcohol 3 in a good yield over two steps (Fig. 4a). Moreover, the conversion of the hydroxyl group on C-3 position of the triol 2z into an imide was accomplished via Mitsunobu reaction after protecting 1,2-diol with the formation of ketal (Fig. 4b). In addition, a chiral α-hydroxyl ketone 5 was synthesized in highly enantioselective manner starting from the triol 2o by means of Silyl ether-protective group manipulation and Parikh-Doering oxidation (Fig. 4c).

Fig. 4
figure 4

Derivatizations of the dihydroxylation products. a Conversion of the primary hydroxyl to amine. b Conversion of the secondary hydroxyl to imide. c Oxidation of the secondary hydroxyl to ketone. (i) TsCl (1.1 equiv), NEt3 (1.2 equiv), DMAP (5 mol%), DCM, 0 °C, 3 h; (ii) BnNH2 (5 equiv), NEt3 (2.5 equiv), MeOH, 80 °C, overnight; (iii) CH(OMe)2 (3.0 equiv), p-TsOH (10 mol%), acetone, r.t., 2 h; (iv) PhthNH (1.2 equiv), PPh3 (3 equiv), DIAD (3.0 equiv), THF, 0 °C−r.t., 12 h; (v) TBDPSCl (1.05 equiv), NEt3 (2.0 equiv), DMAP (5 mol%), DCM, 0 °C-r.t., 12 h; (vi) SO3•Py (5.0 equiv), NEt3 (7.0 equiv), DMSO, DCM, 0 °C−r.t., 24 h; (vii) TBAF (1.5 equiv), HOAc (20 mol%), THF, 0 °C−r.t., 6 h

Mechanistic studies

Subsequently, some mechanistic investigations were carried out for this Mo-catalyzed anti-dihydroxylation. First, we chose the reaction using the allylic alcohol 1c as the standard reaction for the reaction progress kinetic analysis. The yields of the recovered allylic alcohol 1c, the epoxide intermediate 2c′ and the product 2c were determined after reduction with diphneyl sulfide over the reaction time using 1H-NMR spectroscopy (Fig. 5). We noticed the formation of a certain amount of epoxide in the beginning of the reaction, indicating the dihydroxylation reaction consists of a cascade of epoxidation and the following ring opening.

Fig. 5
figure 5

Kinetic study. Reaction kinetic progress study of the Mo-BHA-catalyzed asymmetric anti-dihydroxylation

Furthermore, we quenched the reaction at 5 min and were able to obtain the epoxide 2c′ in 26% yield and 97%ee, which is slightly higher than the enantioselectivity of final triol product 2c (Fig. 6a). The decrease of the optical purity can be reasoned by either the ligand-mismatch effect in the kinetic resolution of the epoxide via hydrolysis or the imperfect regiocontrol in the ring opening process. To verify this, we employed the racemic epoxide 2c′ as precursor under the standard reaction conditions (Fig. 6b). In the beginning we did observe the kinetic resolution of the epoxide 2c′ in favor of the ring opening of the minor enantiomer of the oxirane formed in the epoxidation step. However, the selectivity was very low, and after 30 min the enantiomeric excess of the triol product 2c diminished to 2%. Performing this reaction in the absence of Mo-catalyst resulted in complete shutdown of the ring opening reaction, excluding the non-catalyzed background hydrolysis or perhydrolysis (Fig. 6c). Next, we conducted the ring opening reactions of the enantioenriched epoxide 2ac′ and its enantiomer employing the BHA ligand L6, both of which were prepared through W-catalyzed asymmetric epoxidation28 (Fig. 6d). Within 1 h a full consumption of both epoxides was observed, providing the products in distinct enantiomeric excesses. Comparison of the absolute configuration of 2ac and 2ac′ indicates a C3-selective ring opening as the major reaction pathway29. This result implies that the main reason for the erosion of the enantioselectivity is the competitive ring opening on the C-2 position, and the two enantiomers of the epoxides undergo the ring opening reaction with distinct regioselectivity under the catalysis of the same chiral Mo-catalyst. Moreover, we employed TBS-protected allylic alcohol 1af as substrate and the dihydroxylation reaction still occurred. However, the product 2af was delivered in racemic form, confirming that the presence of hydroxyl moiety as anchoring group is crucial for obtaining successful result of this Mo-catalyzed anti-dihydroxylation (Fig. 6e). Finally, we conducted the ring opening reaction of the epoxide 2ac′ with water in the presence of both Mo-catalyst and the ligand L6. Surprisingly, no hydrolysis of the epoxide occurred in this case, indicating that H2O2 plays a key role in the ring opening of the epoxides intermediates (Fig. 6f).

Fig. 6
figure 6

Control experiments for the Mo-BHA-catalyzed asymmetric anti-dihydroxylation. a Dihydroxylation quenched at 5 min. b Ring opening of racemic epoxide under catalysis of chiral Mo-complex. c Ring opening reaction in the absence of chiral Mo-complex. d Ring opening of enantioenriched epoxides under the catalysis of chiral Mo-complex. e Dihydroxylation of TBDPS-protected allylic alcohol. f Ring opening with water as nucleophile

Discussion

In conclusion, we developed a Mo-BHA-catalyzed asymmetric anti-dihydroxylation of allylic alcohols, providing an efficient entry to 1,2,3-triols in highly enantioselective manner. Being complementary to the Sharpless dihydroxylation in terms of both diastereo- and regioselectivity, our method is bestowed with the following advantages including the high level of diastereo- and enantiocontrol and the used of environmentally benign hydrogen peroxide as oxidant. The preliminary mechanistic investigations reveal that this anti-dihydroxylation consists of an initial enantioselective epoxidation and the following regio- and diastereoselective ring opening.

Methods

Synthesis and characterization

See Supplementary Methods (general information about chemicals and analytical methods, synthetic procedures, 1H and 13C NMR data and HPLC data), Supplementary Figs. 13 (synthetic procedures), Supplementary Figs. 1050 (1H and 13C NMR spectra) and Supplementary Figs. 5187 and Supplementary Table 1 (HPLC chromatograms).

General procedure

To a stirred solution of MoO2(acac)2 (5 mol%)a and BHA L6 (6 mol%)a in EtOAc b (2 mL) at 40 °Cc was added H2O2 (2.5 equiv, 35% w/w aq.). After stirring for 30 min, the alcohols 1 (0.2 mmol) were added to the suspension and stirred at the same temperature for 24 hd. Then the reaction was quenched with diphenyl sulfide (3.0 equiv). After stirring for 1 h at 50 °C, the solvent was removed under reduced pressure. The residue was purified through column chromatography (silica gel, petroleum ether/ethyl acetate) to afford the desired products 2. a MoO2(acac)2 (10 mol%), BHA L6 (12 mol%): 2w-2ae; b 1,4-dioxane: 2y-2ae; c 30 °C: 2o-r, 2u and 2v; d 4 h: 2o, 2p and 2u, 1.5 h: 2q, 2r and 2v.

Synthetic transformations

Full procedures for synthetic transformations to prepare compounds 35 are available in the Supplementary Methods and Supplementary Figs. 46.

Determination of the absolute configuration

For determination of the absolute configuration of triol products 2r, 2z and 2ad, see Supplementary Fig. 7 and 8. The stereochemistry of all the other products was assigned by assuming a common reaction pathway.

Proposed catalytic cycle

A plausible catalytic cycle is proposed in Supplementary Fig. 9.