Diastereoselective Synthesis of Aryl C-Glycosides from Glycosyl Esters via C–O Bond Homolysis

Graphical Abstract

A method to synthesize C-aryl and heteroaryl glycosides from glycosyl esters. The reaction is particularly useful for preparing β-aryl nucleoside analogues, leveraging the photoredox activation of dihydropyridine and decarboxylation to homolytically cleave C–O bonds and afford glycosyl radicals.

C-aryl glycosyl compounds offer better in vivo stability relative to O- and N-glycoside analogues. C-aryl glycosides are extensively investigated as drug candidates and applied to chemical biology studies. Previously, C-aryl glycosides are derived from lactones, glycals, glycosyl stannanes, and halides, via methods displaying various limitations with respect to the scope, functional group compatibility, and practicality. Challenges remain in the synthesis of C-aryl nucleosides and 2-deoxysugars from easily accessible carbohydrate precursors. Herein, we report a cross-coupling method to prepare C-aryl and heteroaryl glycosides, including nucleosides and 2-deoxysugars, from glycosyl esters and bromoarenes. Activation of the carbohydrate substrates leverages dihydropyridine (DHP) as an activating group followed by decarboxylation to generate a glycosyl radical via C–O bond homolysis. This strategy represents a new means to activate alcohols as a cross-coupling partner. The convenient preparation of glycosyl esters and their stability exemplifies the potential of this method in medicinal chemistry.

C-aryl glycosyl compounds are prevalent drug candidates, since the C-glycosidic linkage confers in vivo stability through resistance to hydrolysis and enzymatic degradation.^[1] For instance, C-aryl nucleosides, such as tiazofurin 1,^[2] are prototypical antiviral compounds as they can be recognized by cellular or viral polymerases,^[3] while modifications to their structure lead to disruption and/or termination of replication.^[4] C-aryl nucleosides, including 1^[5] and benzamide C-ribose 2,^[6] show antiproliferative activity (Scheme 1A). Moreover, synthetic C-aryl nucleoside analogues, including 3, are essential to studying the origins of mutagenicity and understanding the mechanism of replication and evolution.^[7] Expanded genetic codes, including dNaM (X) 4, provide a platform for creating therapeutic proteins.^[8] Since nucleosides have high solubility, cell permeability, and in vivo stability, replacing nucleobases with organic fluorophores creates fluorescent tags that can be readily assembled on a DNA synthesizer and applied for bioimaging.^[9]

Scheme 1.

(A) Selected C-Aryl Nucleosides of Medicinal and Chemical Biology Significance and (B) Synthetic Strategies.

Efficient and sustainable synthesis of C-aryl glycosides would facilitate drug discovery and chemical biology studies (Scheme 1B).^[10] Current methods exhibit various limitations. The Friedel–Crafts type glycosylation of electron-rich arenes offers a simple synthetic solution, but lacks the general regiochemical control on the arene.^[11] The nucleophilic addition of aryllithium reagents to lactones precludes many electrophilic and acidic functional groups.^[12] Arylation of glycals via the Heck reaction entails multiple steps from the native sugars.^[13] Stereospecific cross-coupling of glycosyl stannanes offers excellent stereocontrol, but requires pre-generation of the tin reagents with defined stereochemistry and has not been applied to furanoses.^[14] Cross-coupling of glycosyl halides with organometalloarenes,^[15] aromatic nucleophiles,^[16] or aromatic electrophiles^[17] achieved success with a range of carbohydrates, but 2-deoxy sugar substrates, especially 2-deoxyriboses, are susceptible to decomposition, due to the rapid elimination of glycosyl chlorides and bromides to form glycals.^[18] Thus, many gylcosyl halides often require in situ generation.^[19] The Diels-Alder reaction is useful for preparing pyranoses, but cannot be applied to furanoses.^[20] A photoredox decarboxylation reaction leverages an unnatural tetrahydrofuran derivative with a carboxylic acid group at the C1 position, and therefore lacks a general scope of sugar substrates.^[21] Other photoredox coupling reactions based on redox auxiliaries serve as excellent approaches to non-classical C4 or C5-substituted sugars, but do not offer substitution at the anomeric position.^[22]

Carbohydrates free of anomeric protection or glycosyl esters serve as the ideal precursors to C-glycosides, but no C-aryl glycosylation reactions have been reported based on cross-coupling of free carbohydrates or glycosyl esters to our knowledge. Herein, we describe a method to synthesize C-aryl and heteroaryl glycosides, including β-nucleosides and 2-deoxysugars, via the cross-coupling of aryl and heteroaryl bromides with a redox-active glycosyl ester. Upon photoredox activation and electron transfer, the anomeric C–O bond undergoes homolysis and generates a glycosyl radical intermediate. The convenient access to the glycosyl ester and its stability exemplifies the potential of this method in medicinal chemistry.

Redox auxiliaries have emerged to transform hydroxyl groups into viable leaving groups via C–O bond homolysis to form radical intermediates (Scheme 2A). Typical auxiliaries include xanthates,^[23] phosphites,^[24] and oxalates^[25] activated through electron transfer or ight irradiation (Scheme 2A). These methods, however, have not been applied to glycosylation, possibly due to the instability of the corresponding glycosyl esters.^[26] Dihydropyridine (DHP) has emerged as an versatile activating group to produce carbon radicals and carbamoyl radicals, from aldehydes and amines, respectively, via C–C bond homolysis (Schemes 2B and 2C).^[27] The former has been applied to synthesize unconventional glycosides via C–C bond cleavage at the C4 or C5 position (Scheme 2B).^[20] We hypothesize that DHP-mediated radical formation could be combined with subsequent decarboxylation of an alkoxycarbonyl radical to activate the anomeric C–O bond (Scheme 2D). This mechanism represents a new strategy for homolytic activation of hydroxyl groups and formation of radicals.

Scheme 2.

Comparison of radical formation strategies based on photoredox activation of DHP.

We envision that ester 5 is readily available from the condensation of carbohydrates to DHP carboxylic acid (Scheme 3A). Oxidation of 5 by an excited photosensitizer (PC*), followed by deprotonation, can afford radical 6.^[25c] Subsequent fragmentation is driven by the formation of Hantzsch pyridine 8 and generates an alkoxycarbonyl radical 7. Upon ejection of CO₂, 7 can be transformed to glycosyl radical 9,^[28] which enters the catalytic cycle mediated by nickel and cross-couples with aryl bromides. Coordination of 9 to Ni(II) 10 affords Ni(III) 11, followed by reductive elimination to give C-aryl glycoside 14. The resulting Ni(I) 12 is readily reduced to Ni(0) by [PC]^-•.^[29] The initial formation of Ni(0) 13 from a Ni(II) precatalyst may be accomplished by reduction with [PC]^-•, as well.

Scheme 3.

(A) Proposed Mechanism for C-Aryl Glycosylation Based on Dihydropyridine (DHP) Derived Esters and (B) Optimized Conditions.

Catalyst development focused on bis-acetonide protected D-mannofuranose as a model substrate (Scheme 3B). A DHP auxiliary 16 is readily available by condensing aminocrotonate with glyoxylic acid.^[30] Coupling DHP acid 16 with the bis-acetonide protected D-mannofuranose 15 affords O-mannofuranosyl ester 17 in 90% yield. We applied a modified variant of the photoredox-nickel dual catalytic condition for carbamoyl radical generation to couple 17 with PhBr, using 4CzIPN as the photosensitizer,^[25c,31] NiBr₂•DME (DME = 1,2-dimethoxyethane) as the catalyst,^[32] and bipyridine (bpy) as the ligand. The desired C-phenyl-1-deoxy D-mannofuranose 18 was obtained in 81% isolated yield under blue light (467 nm) irradiation at 84 ºC, a temperature required to facilitate DHP fragmentation and the subsequent decarboxylation. A lower temperature led to formation of an ester byproduct, derived from the coupling of 7 to PhBr. Slight excess of 17 ensures a high yield of 18; Using equal molarity of 17 and PhBr gave 18 in 74% yield. The α and β anomers of 17 were separately subjected to the conditions, and both transformed to 18 in comparable yields and high α–selectivity. The stereochemistry was assigned based on NOESY, COSY, and HSQC experiments and compared with literature reports.

We applied the optimized conditions to couple O-mannofuranosyl ester 17 with a range of aryl bromides (Table 1). Due to fluctuations caused by light irradiation, we recorded the actual temperature of the oil bath for each reaction. In general, electron-deficient aryl bromides gave excellent yields of the corresponding C-glycosyl arenes (19-23, 27–29). Performing the synthesis of 17 on a 1.94-gram-scale afforded 21 in 82% isolated yield. Electron-rich electrophiles, including para-methyl and para-methoxy phenyl bromides also gave good to excellent yields (24-25). Para-dimethylaminophenyl bromide, however, is unsuccessful in generating 26. Coupling of 17 with heterocycles, including thiophene with electron-withdrawing substituents, pyrimidine, and pyridines, proceeded to afford the corresponding C-heteroaryl glycosides (32–37) in modest yields. All reactions favor formation of the α-anomer, since the concave β-face is sterically protected by C2, C3, and C4 substituents. The limitation with some electron-rich arenes (26) and heterocycles, such as furan and pyrrole (28 and 29), could be attributed to the incompatibility of these easily oxidizable substrates with the excited photocatalyst.

Table 1.

Scope of C-Aryl Glycosylation with DHP derived Glycosyl Esters^[a]

^[a]Isolated yields. Reaction conditions: ArBr, 0.20 mmol; glycosyl ester, 0.24 mmol (1.2 equiv).

^[b]Reaction performed with 1.94 grams of 17.

^[c]2,2’-Bipyridine-4,4’-dicarboxylic acid as the ligand.

^[d]NiBr₂•DME (10 mol%), bpy (14 mol%).

^[e]4,4’-Dimethoxy-2,2’-bipyridine as the ligand.

We explored the scope of furanoses with various protecting groups.^[33] D-Ribofuranoses containing common protecting groups, such as benzyl, silyl, and benzoyl, underwent smooth C-arylation, forming 38-40 with excellent selectivity for the β-anomer. Immediate application features the synthesis of nucleoside analogues 41-44, precursors to pharmaceuticals 1-2 and unnatural nucleosides 3-4. The low yield of 42 may be attributed to the decomposition of thiazole at the elevated temperature.

D-Xylofuranose and D-glucofuranose underwent β-arylation to afford 45-46. The stereoselectivity was altered to favor the α-anomer for D-galactofuranose 47 and D-arabinofuranose 48, reflecting a dominating effect of the C2 substituent on the stereochemical outcome. Arylation of the benzyl protected 2-deoxy-D-ribose generated a mixture of α and β anomers of 49–51 in a ratio of 2.1:1, presumably due to the lack of steric strains to distinguish the β face from the α face. β-2-deoxyriboses 50-51 are unnatural nucleoside analogues crucial to chemical biology studies. Although the yields for β-50 and β-51 are low, the overall efficiency could still compete with previous multiple-step syntheses based on the Heck reaction and addition of lithium reagents to lactones.^[7a] An alternative approach to 50-51 involves the preparation of their β-C-aryl ribose analogues, which are readily available in high selectivity based on this cross-coupling reaction, followed by reduction of the 2-hydroxyl group.^[34]

A variety of pyranoses, including D-mannopyranose, 2-deoxy-D-glucopyranose, L-rhamnopyranose, and 2-deoxy-D-ribopyranose, proceeded to form C-aryl glycosides 52-58 under the standard conditions. Acetyl protected D-mannopyranose gave low yield of 52, due to facile β-elimination to afford glycals as the by-product. The ¹C₄ conformation of 53-55 was confirmed by the NOE between H1 and H6,7. The observed α-selectivity can be attributed to the kinetic anomeric effect.^[35,36] The mannopyranosyl radical intermediate adapts a chair-like conformation 61,^[37] stabilized by the hyperconjugation of the nonbonding orbital of the ring oxygen, the radical orbital (SOMO), and the σ* orbital of the adjacent C2–O bond. The attack of the nickel intermediate to the mannopyranosyl radical favors the axial position, because the transition state can be stabilized by the donation of the nonbonding electron pairs on the ring oxygen to the antibonding orbital (σ^*‡) of the newly formed C1–Ni σ-bond. In addition, the approach of nickel from the α-face avoids the steric hindrance in the β-face created by the C2-substituent. Benzyl and methyl-protected glucopyranoses display poor reactivity to afford glucopyranosides 58-59 in low yields and as a mixture of α and β-anomers. Product 57 is a precursor to dapagliflozin, a treatment for type II diabetes.^[38] The glucopyranosyl radical intermediate prefers to accommodate a boat conformer 62.^[30] The poor selectivity can be attributed to the contradictory preferences by the steric and the stereoelectronic effect. The steric hindrance at C2 favors the β–attack, whereas the transition state for the α-attack can be stabilized by the kinetic anomeric effect. For a similar reason, D-galactopyranose 60 also displayed unsatisfactory yield and selectivity.

The generality of this deoxygenative coupling method implies applications in complex molecule synthesis via late-state functionalization. We applied the method to derivatizing natural product (+)-sclareolide (Scheme 4). The DHP ester 63 underwent cross-coupling with para-bromobenzoate to afford 64 in 54% yield and 8:1 d.r.

Scheme 4.

Derivatization of (+)-sclareolide from deoxygenative cross-coupling.

In summary, we developed a deoxygenative coupling reaction to prepare C-aryl and heteroaryl glycosides by activating the anomeric hydroxyl groups using a DHP redox auxiliary. Upon light-induced electron-transfer, the glycosyl ester can fragmentize and eject CO₂ to generate a glycosyl radical, which cross-couples with aryl and heteroaryl bromides in the presence of nickel catalysts. The innovation centers at the combination of DHP with decarboxylation as a new means to induce C–O bond homolysis and form radicals. This method overcomes several limitations in existing glycosylation reactions for preparing furanoses and pyranoses. The reaction is particularly useful in synthesizing unnatural β-nucleosides that are of pharmaceutical relevance and chemical biology importance, due to the accessibility, stability and availability of the glycosyl DHP ester precursor. Limitations remain in coupling of some easily oxidizable electron-rich arenes and heteroarenes, in addition to glycosylation of D-glucopyranoses.