Selective Axial-to-Equatorial Epimerization of Carbohydrates

Abstract

Radical-mediated transformations have emerged as powerful methods for the synthesis of rare and unnatural branched, deoxygenated, and isomeric sugars. Here, we describe a radical-mediated axial-to-equatorial alcohol epimerization method to transform abundant glycans into rare isomers. The method delivers highly predictable and selective reaction outcomes that are complementary to other sugar isomerization methods. The synthetic utility of isomer interconversion is showcased through expedient glycan synthesis, including one-step glycodiversification. Mechanistic studies reveal that both site- and diastereoselectivities are achieved by highly selective H atom abstraction of equatorially disposed α-hydroxy C–H bonds.

Graphical Abstract

INTRODUCTION

Rare and unnatural carbohydrates play an essential role in the potency and selectivity of hundreds of glycosylated natural products and pharmaceutical compounds.^1–5 These glycans feature unusual relative/absolute stereochemistry, heteroatom substitution, ring branching, and/or varying degrees of deoxygenation (Figure 1A). Despite biological studies implicating rare and unnatural sugars as significant pharmacophores, synthetic challenges limit access to these important scaffolds.^6,7 Current synthetic strategies require multiple chemical steps and rely on protecting groups to achieve site- and diastereoselective reaction outcomes.^8–10 Concise, selective methods are needed for the expedient synthesis of rare and unnatural sugars.^11–16

Figure 1.

Radical methods for rare and unnatural sugar synthesis. (A) Biomass-derived sugars represent ideal starting materials for the synthesis of rare sugars. (B) This work: selective axial-to-equatorial epimerization of sugars by sequential H atom abstraction (HAA) and H atom donation (HAD).

Radical intermediates have been leveraged by both chemists and nature to transform carbohydrate building blocks into diverse target glycans.^17–20 Radical reactions provide access to a complementary reactivity and selectivity profile compared with polar methods and enable selective C–H bond scission in the presence of O–H and N–H bonds. While radical-mediated transformations have featured prominently in the history of carbohydrate synthesis, challenges attendant to radical generation prevented the full realization of this powerful synthetic strategy.^21–26 The emergence of photochemical tools that enable selective radical formation has led to the development of a suite of selective, catalytic methods for the synthesis of branched, deoxygenated, and isomeric sugars.^27–33 In practice, these new methods enable divergent, one-step access to a broad scope of carbohydrate scaffolds from a small pool of minimally protected feedstock sugars (e.g., d-glucose, d-mannose, d-galactose).

Our group has targeted the development of site- and diastereoselective epimerization methods that enable the interconversion of glycan stereoisomers. Previous conditions provided access to rare and unnatural sugar isomers from biomass starting materials via equatorial-to-axial alcohol isomerization (Figure 1B, gray path).²⁹ This reaction was found to proceed through sequential H atom abstraction (HAA) and H atom donation (HAD) steps mediated by two distinct catalysts. Here, we report conditions to achieve axial-to-equatorial alcohol isomerization by employing a compositionally distinct catalytic system (Figure 1B, blue path). The method provides predictable and selective access to a complementary set of rare sugar targets that cannot be obtained using existing isomerization methods.^34–37

RESULTS AND DISCUSSION

We selected d-α-methylgalactoside 1 as a model substrate for the proposed axial-to-equatorial epimerization. We envisioned that a sterically bulky, electrophilic HAA reagent would promote C4-selective epimerization of the model substrate due to the greater steric accessibility of the equatorially disposed C–H bond compared with axial C–H bonds.³⁸ The formation of C4-equatorial isomer, glucoside 2, was observed in 55% yield by employing catalytic quantities of (Bu₄N)₄W₁₀O₃₂ (TBADT, 1 mol %), 4,4′-dimethoxy diphenyl disulfide (10 mol %), and Bu₄NOP(O)(OBu)₂ (20 mol %) in MeCN at room temperature under near-UV (390 nm) LED irradiation (Figure 2A). No reaction was observed to occur in the absence of TBADT, disulfide, or light; the reaction yield was diminished in the absence of Bu₄NOP(O)(OBu)₂. Other diaryl disulfides and thiophenols were also found to be effective as H atom donors (see the Supporting Information for full reaction optimization details). MeCN, acetone, and their aqueous mixtures (up to 15% H₂O) were all found to be suitable solvents for the reaction.

Figure 2.

Reaction development. (A) Optimized reaction conditions for the epimerization of minimally protected pyranosides. (B) Optimized reaction conditions for the epimerization of unprotected pyranoses. Reactions were conducted at a 1.0 mmol scale in duplicate and the average ¹H NMR yield is reported (nitrobenzene or DMF used as an external standard). See the Supporting Information for more details.

The reaction conditions were reoptimized for unprotected sugar substrates, such as d-fructose 3, which exhibited diminished reactivity under TBADT conditions due to their limited solubility in organic solvents. Using aqueous conditions employing catalytic quantities of Na₄W₁₀O₃₂ (NaDT, 1 mol %), 4,4′-diaminodiphenyl disulfide (10 mol %), and NaOP-(O)(OBu)₂ (20 mol %) in 1:1 acetone/H₂O, d-fructose 3 reacted to form l-sorbose 4 in 54% yield (Figure 2B).³⁹ The H atom donor catalysts, 4,4′-diaminodiphenyl disulfide and the corresponding 4-amino thiophenol, were found to be uniquely suited for the NaDT conditions, as nearly all other disulfides and thiols we evaluated provided diminished yields (see the Supporting Information).

A range of monosaccharides and glycans was evaluated as substrates under either TBADT- or NaDT-catalyzed conditions to assess the scope and selectivity of the reaction (Figure 3). α-Configured fucosides 5 and 7 each reacts cleanly at C4 to afford quinovoside products 6 and 8, respectively. Quinovoside 6 is also obtained from the reaction of l-α-methyl rhamnoside 9, in this case, arising from epimerization at C2 rather than at C4. To showcase the synthetic utility of this epimerization method, we developed a scalable synthesis of the rare sugar l-quinovose and its derivatives from rhamnoside 9, the more accessible and economical 6-deoxy l-sugar (Figure 3). While 9 reacts to form 6 in 38% yield with minimal anomerization (52:1 α/β under TBADT conditions, the total yield of 6 increased to 54% (7:1 α/β) when the reaction solvent was switched from MeCN to 15% H₂O in acetone. Due to ultimate cleavage of methyl aglycone, the higher-yielding conditions were carried forward despite the increased formation of the β anomer to access l-quinovose 10 and trimethyl-l-quinovose 11 in 86 and 88% yields, respectively. Our synthesis of 11—a bioactive rare sugar found in xanthone-type natural product, calixanthomycin A—cuts 3 steps from the previous synthesis of 11, which proceeds from rare l-glucose (> $7,000 per mol) and requires redox manipulation and protecting group interconversions (13% yield overall).^40,41 Stereochemistry adjustment, rather than oxidation state adjustment, allows our synthesis to leverage a biomass-derived starting material (< $50 per mol).⁴¹

Figure 3.

Synthetic scope of DT-catalyzed epimerization. Isolated yields are the average of two runs; numbers in parentheses are ¹H NMR yields (average of two runs) with nitrobenzene or DMF as an external standard. Standard reaction conditions for DT-catalyzed epimerization: 1.0 mmol scale, 1 mol % DT, 10 mol % Ar₂S₂, 20 mol % ⁻OP(O)(OBu)₂ (Bu₄N⁺ or Na⁺), 0.2 M in MeCN or acetone/H₂O, room temperature, and near-UV (390 nm) LED. Reaction conditions (time, solvent, concentration) were modified for select substrates to achieve optimal yields and selectivities. Conditions: (a) 0.05 M substrate in 3 M HCl, 80 °C, and 4–5 h; (b) 9 equiv MeI, 9 equiv NaOH, DMSO, RT, and 15 min; and (c) 2 mol % Sc(OTf)₃, 20 equiv Ac₂O, 45 °C, and 18 h. See the Supporting Information for full experimental details. ^aSee ref 41.

We hypothesized that the isomerization of 1,6-anhydrosugars, in which a ring flip enforces an axial disposition of substituents that would otherwise be equatorial, could provide access to idose derivatives. Indeed, the reaction of d-anhydromannose 12 affords d-anhydroidose 13 in 60% yield via sequential epimerization at both C3 and C4 positions. d-anhydroidose 13 was also obtained from the reaction of d-anhydroglucose 14; in this case, three sequential stereocenter inversions occur, affording 13 in 47% yield. Subsequent ring-opening acetolysis of 13 reveals d-idose pentaacetate 15 in nearly quantitative yield. The presence of an acetamide substituent in N-acetyl-anhydroglucosamine 16 shuts down epimerization at C2, and N-acetyl-anhydrogulosamine 17 was obtained from the reaction of 16 (0.5 g isolated, 53% yield). The structure of 17 was confirmed by X-ray crystallography. Previous synthetic routes to N-acetyl-gulosamine require six steps from d-xylose (8–15% yield).⁴²

Fully unprotected sugars also undergo selective epimerization under NaDT conditions. For example, l-quinovose 10 (29% yield) and l-xylose 20 (33% yield) were obtained from l-fucose 18 and d-arabinose 19, respectively. Importantly, product 20 (isolated as the diacetonide) remains enantiopure, suggesting that efficient epimerization does not occur from the minor conformer. d-Lactulose 21 reacts exclusively at the pyranose ring to afford d-celliobiulose 22.

The optimal reaction conditions were extended to inositols, another important class of cyclic polyols. Scyllo-inositol 24, a molecule currently under investigation for the prevention of Alzheimer’s disease, was obtained selectively from the reaction of myo-inositol 23 in 54% yield.^43–46 The reaction conditions were readily amenable to scale, and we were able to isolate 1.4 g of 24 without the need for chromatography. Our route improves upon previous one-step chemical (20% yield) and enzymatic (55% yield) syntheses of 24 from 23 while avoiding the overreliance of protecting groups present in other synthetic methods (64% yield over nine steps from 23).^47–51 Pinitol 25 reacts to form 1-OMe-scyllo-inositol 26 in 62% yield via double epimerization; multistep synthetic sequences are often required for the synthesis of monosubstituted inositols.⁵² Orthoformate-protected myo-inositol 27 undergoes selective epimerization at just one position, forming protected epi-inositol 28 in 71% yield.

The site- and diastereoselectivities observed under DT-promoted epimerization conditions are completely complementary to those obtained using previously reported quinuclidine-promoted conditions, reflecting the unique selectivity preferences of the two distinct HAA reagents. For example, under quinuclidine-promoted conditions, l-β-methylfucoside 29 reacts to form a C2 epimer, l-β-methyltaloside 30, in 50% yield (Figure 4, top).²⁹ In contrast, under TBADT-promoted conditions, 29 reacts to form a C4 epimer, l-β-methylquinovoside 31, in 46% yield. This complementarity was perfectly preserved when more complex fucosides possessing distinct aglycones were evaluated as substrates. Under quinuclidine-promoted conditions, steroidal fucoside 32 and disaccharide 35 undergo C2-selective, equatorial-to-axial epimerization to provide l-taloside products 33 and 36, respectively; using DT-promoted conditions, 32 and 35 undergo C4-selective, axial-to-equatorial epimerization to provide l-quinovosides 34 and 37, respectively. To assess whether site- and diastereoselectivity differences extended to other complex unprotected glycans, the trisaccharide d-raffinose 38 was evaluated as a substrate. Under quinuclidine-catalyzed reaction conditions, 38 reacts to form the allose-containing trisaccharide 39, while under NaDT-promoted conditions, d-theanderose 40 is obtained in 20% isolated yield.²⁹ These findings provide an early illustration of the feasibility of using tailored isomerization reactions for the interconversion of complex glycans. We anticipate that these methods can provide a complementary strategy for “glyco-randomization” of glycosylated natural products and oligosaccharides, which circumvents the laborious process of generating glycoside libraries through iterative glycosylation reactions.^53–55

Figure 4.

Complementary site- and diastereoselective epimerization of saccharides enable rapid glycan assembly. Isolated yields are the average of two runs; numbers in parentheses are ¹H NMR yields (average of two runs) with nitrobenzene or DMF as an external standard. Standard reaction conditions for quinuclidine-catalyzed epimerization: 0.2 mmol scale, 10 mol % quinuclidine, 50 mol % adamantane thiol, 25 mol % Bu₄NOBz(4-Cl), 0.2 M MeCN/DMSO, room temperature, and blue LED. Standard reaction conditions for DT-catalyzed epimerization: 0.2 mmol scale, 1 mol % DT, 10 mol % Ar₂S₂, 20 mol % ⁻OP(O)(OBu)₂ (Bu₄N⁺ or Na⁺), 0.2 M in MeCN or acetone/H₂O, room temperature, and near-UV (390 nm) LED. Reaction conditions (time, solvent, concentration) were modified for select substrates to achieve optimal yields and selectivities. See the Supporting Information for full experimental details. ^aSee ref 29.

We next sought to explore the strategic implications of glycan isomer interconversion for the synthesis of O-functionalized substrates. The selective functionalization of carbohydrate O–H bonds is a central challenge in the glycosciences due to the importance, and difficulty, of forging selective glycosidic linkages and the numerous rare and unnatural sugars that contain functionalized O–H bonds.⁵⁶ Transformations that leverage the selective coordination and activation of 1,2-cis diol motifs have emerged as a powerful and predictable strategy to achieve highly site-selective O–H modification within sugar scaffolds.^56–60 However, similarly robust, selective, and predictable methods for promoting reactions at other positions are rare.^61–64 We envisioned that our epimerization method could be combined with site-selective O-functionalization tools to expand the scope of targets that can be accessed using these cis-diolate methodologies. For example, Taylor has reported the direct, site-selective arylation of carbohydrates under copper-mediated Chan–Lam cross-coupling conditions.⁶⁵ The method enables the efficient synthesis of O-arylated pyranosides, such as 41, through the formation of a boronic ester cis-diolate intermediate. The reaction of 41 under TBADT conditions predictably affords the C4 epimer, 42, in 84% yield. Notably, the direct 2-O-arylation of glucose-configured substrates, such as 42, is not feasible using these conditions due to the lack of the mechanistically necessary 1,2-cis diol motif; related 2-O-substituted glucosides are difficult to access by conventional methodologies (c.f. seven steps to access 2-O-methyl quinovoside 43).⁶⁶

Site-selective reagent-controlled glycosylation is a powerful way to forge certain glycosidic linkages. Employing diarylborinic acid-catalyzed Koenigs–Knorr-type conditions developed by Taylor, disaccharide 44 can be synthesized in a single synthetic step.⁶⁷ The reaction of 44 under TBADT conditions affords the doubly epimerized product, 45, access to which would otherwise require a protecting group-based synthetic scheme. Collectively, these examples demonstrate the strategic integration of isomerization tools with site-selective O–H functionalization methods for rapid glycan assembly.

We performed a series of experiments to investigate the basis for the site- and diastereoselectivities of the reaction. Drawing from previous mechanistic studies of related reactions, we propose a plausible mechanism involving sequential H atom abstraction and H atom donation mediated by excited-state DT (W=O*) and aryl thiol (Ar-SH), respectively (Figure 5A).^34,68–70 Resubjecting α-methylglucoside 2 to the reaction under H/D isotope exchange conditions did not result in the formation of α-methylgalactoside 1, and recovered α-methylglucoside 2 (95% yield) contained only trace detectable d-incorporation after 8 h (Figure 5B).⁷¹ This experiment reveals that 2 does not undergo H atom abstraction and implicates the HAA step as selectivity-determining. When the same isotope exchange experiment was performed using α-methylgalactoside 1, α-methylglucoside 2 was observed in 44% yield after 6h, along with 36% recovered 1 (Figure 5C). Significant deuterium incorporation was detected at the C4 position of both product 2 and recovered 1 (16 and 19%, respectively), indicating unselective H atom donation to the radical intermediate.

Figure 5.

Proposed mechanism and mechanistic studies. (A) Proposed mechanistic picture for epimerization. (B) Deuterium labeling studies indicate that the reaction product does not react under standard reaction conditions. (C) Deuterium labeling studies indicate that the radical intermediate is quenched by unselective H atom donation, forming starting material and product. (D) Site-selective isotopic labeling of glucosides (R = TBS). Conditions: 0.2 mmol scale, 1 mol % TBADT, 10 mol % (4-OMe-C₆H₄)S₂, 20 mol % Bu₄NOP(O)(OBu)₂, 0.2 M solvent, and near-UV LED; B, d₃-MeCN, 8 h; C, d₃-MeCN, 6 h; and D, 15% D₂O in d₆-acetone, 8 h. ¹H NMR yields with nitrobenzene as an external standard; d-incorporation quantified by ¹H NMR of purified substrates. See the Supporting Information for full experimental details.

Collectively, these experiments support a kinetically controlled mechanism in which the overall diastereoselectivity of the reaction is determined by the initial H atom abstraction step. Importantly, the mechanism through which diastereoselectivity is achieved reveals the basis for predictable site selectivity: radical generation can occur only at positions possessing an axially disposed hydroxyl group (i.e., an equatorially disposed C–H bond). To test this hypothesis, we carried out the epimerization of mannoside (C2 axial), alloside (C3 axial), and galactoside (C4 axial) substrates under isotope exchange conditions (Figure 5D). Consistent with our predictive model, deuterium incorporation into the sugar scaffold is exclusively observed at the position of the axial hydroxyl substituent in the starting substrate (41–56% yield, 80–82% d-incorporation).⁷² We anticipate that this direct and selective d-incorporation strategy will enable rapid access to diverse sugar isotopologs of potential interest to the glycoscience community.^73–76

CONCLUSIONS

In summary, we have identified a radical epimerization reaction that enables selective access to diverse rare sugars and cyclic polyols from minimally protected precursor glycans. The method features predictable reaction outcomes that are complementary to other sugar isomerization methods. Both the site- and diastereoselectivity of the transformation were found to be achieved by selective H atom abstraction of equatorially disposed α-hydroxy C–H bond(s). This work further demonstrates the potential of selective interconversion of glycan isomers and its applicability to enable the rapid synthesis of significant rare and unnatural sugar derivatives.