An Asymmetric Stereodivergent Strategy Towards Aminocyclitols

Abstract

A concise asymmetric synthesis of aminocyclitols such as diastereomeric 2-deoxystreptamine analogues and conduramine A is described. The Pd-catalyzed asymmetric desymmetrization of meso 1,4-dibenzolate enables the synthesis of highly oxidized cyclohaxane architectures. These scaffolds can potentially be used to access novel aminoglycoside antibiotics and enantiomerically pure α-glucosidase inhibitors.

Aminocyclitols such as 2-deoxystreptamine (DOS) 1 and conduramine A 2 are important structural motifs found in several aminoglycoside antibiotics and α-glucosidase inhibitors (Figure 1).¹ Although the synthesis of 2-deoxystreptamine has received significant attention², the need for structural replacements continues to be an essential area of research.³ Toxicity is observed when natural 2-deoxystreptamine containing aminoglycosides such as kanamycin 3 and tobramycin 4 are employed in treating bacterial infections.⁴ In addition, the continual emergence of resistant bacterial strains against such aminoglycosides justifies the further exploration of novel antibiotics that are less susceptible to enzymatic inactivation.⁵

Figure 1.

Clinically relevant aminoglycoside antibiotics.

A majority of the strategies towards the synthesis of novel analogues of 3 and 4 have focused on modifying one or more of the aminosugars (Rings I and III in Figure 1). A strategy that has received less attention involves the modification of DOS itself (ring II). The ubiquity, the central location of DOS, and the hydrogen bonds that it forms with 16S RNA may suggest its significance. Further, recent studies on the modification of ring II have resulted in several potent aminoglycosides.^6,7,8

One strategy for modifying ring II that has received virtually no attention, involves the replacement of DOS with its diastereomer.⁹ This modification would have a minimal influence on the overall positive charge density and conformational flexibility of the aminoglycoside, yet, it would significantly modify the three dimensional structure of the molecule. This altered structure may present new opportunities for differential binding to RNA and accssibility to scaffolds that may be less prone to enzymatic modification.

We chose 2-deoxy-4,5,6-tris-epi- (5a) and 2-deoxy-3,4-bis-epi-streptamine (6a), the former since it retains the symmetry and the latter because it does not (Figure 2). Similar to 2-DOS, diamine 5a possesses a plane of symmetry, however, when incorporated into an aminoglycoside, it becomes a chiral motif. Thus, a synthetic equivalent must posses a differentiation element that allows for the asymmetric incorporation as envisioned in 5b (Figure 3).

Figure 2.

Diastereomeric 2-deoxystreptamine targets.

Figure 3.

Stereodivergent strategy towards 5b and 6b.

To efficiently access and investigate the biological properties of aminoglycosides that are diastereomeric at ring II, differentially protected diastereomeric DOS analogues must be readily accessible. The challenge associated with a selective synthesis of such analogs lies in the presence of five highly oxidized contiguous stereocenters. Although, several racemic syntheses of diastereomeric DOS exist,¹⁰ current methods to access enantiomerically pure diastereomeric DOS analogues require multi-step sequences that employ the use of chiral pool starting materials (sugars).¹¹ Herein we describe an efficient synthesis of differentially protected diastereomeric deoxystreptamines 5b and 6b using asymmetric catalysis where both enantiomers of the desired products can be obtained.

To access DOS analogues 5b and 6b, our goal (Figure 3) was to maximize efficiency by: (1) designing a flexible route that allows for the synthesis of either enantiomer of several aminocyclitols, and (2) employing diastereoselective transformations to fully utilize stereochemical information embedded in the molecule. We have previously described a palladium-catalyzed asymmetric allylic azidization reaction for the desymmetrization of meso dibenzoate 8 as a method for preparing enantiomerically pure amino alcohols (Scheme 1).¹² We envisioned that employing such a strategy would provide an efficient entry into various aminocyclitols that could be later functionalized.

Scheme 1.

Desymmetrization of dibenzoate 8.

Retrosynthetically, diastereomeric DOS 5b and 6b can be obtained via the trans- and cis-dihydroxylation of a common intermediate 7, respectively (Figure 3). Initially two strategies were examined to incorporate the second nitrogen in analog 5b as outlined in Scheme 2. First, the depicted primary carbamate was subjected to the Rh-catalyzed intramolecular nitrogen insertion;¹³ however, this resulted in the exclusive formation of the corresponding ketone. Second, when the tosyloxime was subjected to base, instead of the α-aminoketone (Neber rearrangement), we obtained aromatized products.

Scheme 2.

Initial strategies for second nitrogen incorporation

In an alternative strategy, we envisioned that the second nitrogen of analog 5 could be installed via a nitrosation. Analogously, the second nitrogen of analog 6b would be installed via the regioselective opening of an epoxide. Carbamate 7 could be obtained via the Pd-catalyzed asymmetric desymmetrization of meso dibenzoate 8.

Previous work has shown that a catalyst derived from π-allylpalladium chloride dimer and ligand 9 gave a mixture of the desired product 11 and the regioisomeric azide 12, while a catalyst derived from ligand 10 gave a mixture of azide ent-11 and diazide 13 (Scheme 1).^12c Gratifyingly, the desired carbamate 7 was obtained as the sole product in 87% yield and 97% ee via a one-pot protocol using ligand 9.¹⁴ The low catalyst loadings and low temperature manipulation of the reactive azide for the desymmetrization process and the Staudinger reduction contributed to the exclusive formation of carbamate 7 (Scheme 1).

With a reliable method to obtain intermediate 7 in hand, we proceeded with the hydrolysis of the benzoate group to provide allylic alcohol 14 (Scheme 3). Directed epoxidation of this alcohol 14 afforded epoxide 15 as a single diastereomer whose further oxidation provided epoxyketone 16. Enolization of epoxyketone 16 with potassium tert-butoxide, followed by quenching with iso-amyl nitrite, afforded oxime 17 as a single isomer of undetermined geometry. Reduction of the carbonyl carbon from the least hindered face using sodium borohydride afforded alcohol 18 as a single diastereoisomer.

Scheme 3.

Synthesis of oxime 18.

Reaction Conditions: a) K₂CO₃, MeOH, 93%; b) m-CPBA, CHCl₃, 99%; c) TPAP, NMO, CH₂Cl₂, 86%; d) KOt-Bu, i-amyl nitrite, THF, −78 °C; e) NaBH₄, MeOH, 56% over 2 steps.

The selective reduction of oxime 18 proved to be challenging. Ultimately, we determined that a nickel boride-mediated reduction followed by acylation with acetic anhydride afforded diacetate 19 as a single diastereomer (Scheme 4). The opening of epoxide 19 was attempted with several Lewis and Brönsted acids. Clean trans diaxial opening of epoxide 19 with aqueous trifluoroacetic acid provided an intermediate triol that was immediately trapped with benzyl choloroformate. NMR spectroscopic analyses, which were later corroborated by optical rotation and x-ray studies, showed that under these conditions we had not only opened the epoxide but also unexpectedly deprotected both nitrogens, destroying chirality by forming meso triol 20 which also establishes the stereoselectivity of the oxime reduction.

Scheme 4.

Synthesis of diastereomers 5b and 20.

Reaction Conditions: a) NaBH₄, NiCl₂•6H₂O, MeOH, −78 °C; then Ac₂O, 52%; b) TFA, H₂O (1:1), 110 °C; then BnOCOCl, K₂CO₃, MeOH:H₂O (3:1), 50%; c) BzCl, Et₃N, THF, 70% d) NaBH₄, NiCl₂•6H₂O, MeOH, −78 °C to rt, 88% (e) TFA, H₂O (1:1), 75 °C, 55% (5:1 dr).

The solution to this problem lay in the installation of a more robust differentiating group with the dibenzoylation of oxime 18. Gratifyingly, the reduction of the resulting oxime 21 was greatly facilitated by the incorporation of an electron withdrawing group (Scheme 3). The reduction of oxime 21 progressed with the in situ migration of the benzoyl group from O to N to give epoxide 22 (structure confirmed by NMR and x-ray).¹⁵ When conditions similar to those attempted for the opening of epoxide 19 were attempted, we were pleased to find that the opening of the epoxide proceeded with 5:1 regioselectivity and that the benzoyl group was stable to the reaction conditions. This completed the synthesis of our first diastereomeric DOS core 5b.

In order to evaluate synthetic flexibility enabled by the Pd-catalyzed desymmetrization we pursued the synthesis of diasteromer 6b via an alternative route (Scheme 5). To this end, a diastereoselective osmium tetroxide-catalyzed cis-dihydroxylation of allylic alcohol 14 afforded an intermediate triol that was immediately protected to generate acetonide 23 (Scheme 5). The treatment of alcohol 23 with methanesulfonyl chloride gave mesylate 24. Elimination of mesylate 24 initially proved problematic. Ultimately, high temperature microwave conditions yielded olefin 25. Epoxidation of olefin 25 afforded epoxide 26 as a single diastereomer (structure confirmed by NMR and x-ray).¹⁵ The stereoselectivity of this reaction is controlled by both the steric bulk of the proximal acetonide and, to a larger extent, by the ability of the carbamate in 26 to direct the epoxidation.¹⁶ Racemic epoxide 26 has been previously converted to conduramine A 2; therefore, our synthetic approach constitutes an asymmetric formal synthesis of this essential building block used in the synthesis of α-glycosidase inhibitors (Scheme 5).^17,18

Scheme 5.

Synthesis of diastereomer 6b.

Reaction Conditions: a) OsO₄, NMO, acetone; then 2,2-dimethoxypropane, PPTS, 62% over 2 steps; b) MsCl, DIPEA, 95%, CH₂CI₂; c) DBU, microwave, 150 °C, 55%; d) m-CPBA, CHCl₃, 83%; e) NaN₃, NH₄Cl, n-propanol, reflux; then Boc₂O, 82%.

To complete the synthesis of diastereomeric DOS 6b, the opening of epoxide 26 was attempted. Due to the electron deficient nature of the ring, forcing conditions were required. By refluxing epoxide 26 with sodium azide in n-propanol, the expected trans diaxal addition product was formed, but with the concomitant loss of the tert-butoxycarbonyl group. Trapping the amine by the addition of di-tert-butyldicarbonate afforded the final diastereomer 6b.

In conclusion, we have developed a concise, asymmetric synthesis of two diastereomers of 2-deoxystreptamine and an asymmetric formal synthesis of conduramine A. Our approach demonstrates the utility of a one-pot catalytic asymmetric desymmetrization in the synthesis of epimers of ring II in aminoglycosides. The two synthetic strategies presented are concise and display a range of highly diastereoselective transformations that provide several intermediates that can potentially be of value in the design of a range of unnatural aminocyclitols, such as 2-deoxystreptamine analogs and other conduramines.

Experimental Section

1S,4R- Benzoic acid 4-tert-butoxycarbonylamino-cyclohex-2-enyl ester

To a flame-dried round-bottomed flask was added [(η³-C₃H₅)PdCl]₂ (29.0 mg, 0.079 mmol), (S,S)-9 (164 mg, 0.2376 mmol), and dibenzoate 8 (12.5 g, 39.62 mmol). The flask was the placed under reduced pressure (vacuum pump) for 10 seconds and refilled with Ar; this purging procedure was repeated three times to ensure no oxygen remained in the reaction vessel. After being placed in an Ar atmosphere, degassed THF (17 mL) was added and the mixture was stirred for 10 minutes at room temperature. Freshly distilled azidotrimethylsilane (6.31 mL, 47.54 mmol) was added dropwise at 0 °C and the reaction was stirred at this temperature for 1.5 hours. Water (30 ml) was added to the reaction mixture followed by a dropwise addition of solution of trimethyl phosphine (100 mL of a 1M solution in THF) over 2 hours. Upon complete disappearance of the intermediate allylic azide, triethylamine (13.0 mL) and di-tert-butyl-dicarbonate (13.3 g, 66.66 mmol) were added and the reaction was stirred for 12 hours at room temperature. The reaction was diluted with diethyl ether (300 ml) and washed with saturated sodium bicarbonate, brine, dried (MgSO₄) and concentrated. Silica gel chromatography using 10% ethyl acetate/hexanes afforded 10.69 g (87%) of the title compound as a clear oil. R_f = 0.5 in 20% ethyl acetate/hexanes; ^[α]D₂₄ = −75.84 (c 0.18, CH₂Cl₂); ¹H (400 MHz, CDCl₃): δ 8.03 (dd, J = 8.0, 1.6 Hz, 2H), 7.54 (dd, J = 8.0, 8.0 Hz, 1H), 7.42 (dd, J = 8.0 Hz, 2H), 5.93 (ddd, J = 10.1, 3.4, 1.0 Hz, 1H), 5.89 (dd, J = 10.1, 2.4 Hz, 1H), 5.42 (m, 1H), 4.60 (bd, J = 7.9 Hz, 1H), 4.19 (m, 1H), 2.01–1.89 (m, 3H), 1.72 (m, 1H), 1.4 (s, 9H) ppm; ¹³C NMR (75 MHz, CDCl₃): δ 166.1, 155.2, 133.8, 133.0, 130.4, 129.6, 128.4, 128.1, 79.6, 67.4, 46.0, 28.4, 26.0 ppm; IR (thin film) νmax 3411, 3315, 3065, 3027, 2943, 2867, 1702, 1530, 1246, 1064 cm⁻¹; HRMS (EI, [MC₁₈H₂₃NO₄ -(C₄H₉)]+) Calc’d for C₁₄H₁₅NO₄: 261.1001 Found: 261.1010.

1S,4R-(5-Hydroxy-7-oxa-bicyclo[4.1.0]hept-2-yl)-carbamic acid tert-butyl ester

To a solution of alcohol 14 (56 mg, 0.26 mmol) in anhydrous chloroform (20 mL) was added m-chloroperoxybenzoic acid (50 mg, 0.29 mmol). The reaction was stirred at room temperature for 24 hours (consumption of starting material was monitored by GC). Upon completion, the reaction mixture was washed with saturated sodium hydrogen carbonate, brine, dried (MgSO₄), and concentrated to afford 60 mg (99%) of title compound as a clear oil that slowly crystallized to a white solid. R_f = 0.7 in 100% ethyl acetate; ^[α]D₂₆ = +22.12 (c 2.2, CH₂Cl₂); ¹H NMR (400 MHz, DMSO-d₆): δ 6.72 (d, J = 8.0 Hz), 4.82 (d, J = 5.3 Hz, 1H), 3.82 (m, 1H), 3.74 (m, 1H), 3.24 (dd, J = 4.0 Hz, 1H), 3.17 (dd, J = 4.0, 2.6 Hz, 1H), 1.48–1.27 (m, 4H), 1.38 (s,1H) ppm; ¹³C NMR (100 MHz, DMSO-d₆): δ 155.1, 77.7, 64.6, 56.1, 54.7, 44.7, 28.2, 26.4, 24.7 ppm; IR (thin film) νmax 3348, 2977, 2938, 2871, 1702, 1506, 1456, 1392, 1367, 1326, 1249, 1170, 1068, 1003, 929, 895, 848, 776 cm⁻¹; HRMS (EI, [M]+) Calc’d for C₁₁H₁₉NO₄: 229.1314 Found: 229.1315. The stereochemistry of the desired product was initially based on hydrogen bond-directed epoxidation and later assigned by X-ray analysis of a downstream intermediate.¹⁵

1S,4R-(5-Oxo-7-oxa-bicyclo[4.1.0]hept-2-yl)-carbamic acid tert-butyl ester

Tetra-n-propylammonium perruthenate (0.66 g, 1.88 mmol) and activated 4°A molecular sieves (19 g) were added to a solution of 15 (8.62 g, 37.62 mmol) in dichloromethane (268 mL). Solid 4-methylmorpholine N-oxide (6.61 g, 56.43 mmol) was added and the slurry was stirred at room temperature for 8 hours. The reaction was poured onto a bed of silica gel and eluted with dichloromethane. The eluent was concentrated and the crude product was purified using silica gel flash chromatography (30% ethyl acetate/hexanes) to afford 7.36 g (86%) of the title compound as a white solid. R_f = 0.25 in 20% ethyl acetate/hexanes; melting point: 86 – 88 °C; ^[α]D₂₃ = +178.0 (c 0.2, CH₂Cl₂); ¹H NMR (400 MHz, CDCl₃): δ 4.86 (d, J = 9 Hz, 1H), 4.25 (q, J = 9 Hz, 1H), 3.65 (d, J = 4 Hz, 1H), 3.29 (d, J = 4 Hz, 1H), 2.53 (ddd, J = 18.0, 4.0, 4.0 Hz, 1H), 2.18 (ddd, J = 18.0, 4.0, 4.0 Hz, 1H), 1.85 (m, 1H), 1.46 (s, 9H) ppm; ¹³C NMR (125 MHz, CDCl₃): δ 203.2, 155.1, 80.2, 57.5, 55.9, 46.3, 34.9, 28.3, 22.9 ppm; IR (thin film) νmax 3331, 2976, 1693, 1524, 1455, 1391, 1365, 1306, 1249, 1167, 1056, 1007, 957, 876 cm⁻¹; HRMS (EI, [MC₁₁H₁₇NO₄-(C₄H₉)]+) Calc’d for C₇H₉NO₄: 171.0532 Found: 171.0539; Elemental analysis Theoretical C58.14, H7.54, N6.16 Found C57.96, H7.54, N6.04.