[Au]/[Ag]-Catalyzed Glycosidation of Ethynylcyclohexyl Glycosyl Carbonates Enables Direct Stereoselective Synthesis of 2-azido-2-deoxy- β-Mannopyranosides

Abstract

We describe the [Au]/[Ag]-catalyzed activation of ethynylcyclohexyl 2-azido-4,6-O-benzylidene mannosyl carbonates for β-mannosamine linkage preparation in high yield and selectivity in a temperature and concentration-dependent manner. VT-NMR studies reveal an anomeric triflate intermediate generated in the absence of acceptor alcohol that is stable up to −10 °C. The generality of the protocol was illustrated by successful application to a series of acceptors and by synthesis of a fully protected Streptococcus pneumoniae type 19F capsular trisaccharide.

Graphical Abstract

Glycosylation is one of the most important reactions that is employed to synthesize complex glycan targets.¹ It can be visualized as the conjugation between a glycan and another molecule termed the aglycon.^1b In glycosylation, the glycan portion usually contains a leaving group that gets activated in the presence of promoters resulting in the formation of a highly electrophilic intermediate; glycans possessing the leaving group are referred as glycosyl donors.¹ Many pioneering efforts over the past century culminated in the development of various glycosyl donors for the formation of diverse glycosidic bonds. Often the highest 1,2-trans stereoselectivities were accomplished by furnishing the glycosyl donor with a participating group such as an ester at the C2 position so that it can impart neighboring group participation to afford 1,2-trans glycosides in a facile manner.¹ Alternatively, participation through thioether auxiliaries^1c and picolinyl groups^1d from C2 or other remote positions has been explored. However, the stereoselective synthesis of 1,2-cis glycosides is still a difficult task. These challenging 1,2-cis glycosides can be synthesized by either the intramolecular aglycon delivery (IAD),^2a–c reactions at low temperatures,^2d H-bond mediated aglycon delivery (HAD),^2e by steric buttressing,^2f or by conversion of a 1,2-trans- to a 1,2-cis glycoside by inversion of configuration at C2.^2g–i

Of the many sugars, mannose occupies a special position as many bacterial, mammalian, and microbial species contain high molecular weight mannans differentiated by 1,2-trans (α-) and 1,2-cis (β-) mannosidic linkages.³ The C2-position of the mannopyranosyl sugar is disposed axially, therefore, the synthesis of 1,2-trans mannosides becomes relatively easier due to the strong anomeric effect as well as the favorable steric hindrance of the axial substituent. The synthesis of 1,2-cis mannosides was reported by intramolecular aglycon delivery,^2a stereochemical editing,^2h Crich β-mannosylation equipping the glycosyl donor with 4,6-benzylidene acetals^2d or HAD.^2e The ManNAc (also called N-acetylmannosamine or 2-acetamido-2-deoxy-mannopyranose) has emerged as a common component in capsular polysaccharides of bacterial, prokaryotes and is a key precursor for the biosynthesis of N-acetylneuraminic acid in mammals.³ Additionally, the occurrence of ManNAc residues in the biologically important glycans,³ asparagine-linked glycopeptides,⁴ and vaccine candidates warrants the development of novel glycosylation methods for the β-stereoselective synthesis of mannosamine derivatives (Scheme 1A). The synthesis of 1,2-trans (α-) mannopyranosides can be accomplished through the neighboring group participation whereas the stereoselective synthesis of 1,2-cis (β-) glycosides of ManNAc is cumbersome. Earlier studies showed that glycosyl donors that produced excellent β-mannosylation met with limited success when applied for the synthesis of glycosides of β-mannosamine.^5a As a result, a vast majority of literature reports resorted to indirect methods of converting glucose to mannose post-glycosylation event. The stereochemical editing of βGlcp→βManpNAc demands careful planning of other functional groups (Scheme 1B). In this context, the direct stereoselective synthesis of β-glycosides of mannosamines is of paramount significance. Notably, complete β-manno selectivity was noticed while synthesizing a trisaccharide from a thiophenyl disaccharide donor equipped with 2-azido mannoside.^5b The β-manno selectivity dropped from 1:10 (α:β) for C2-OBn donors to 1:4.9 (α:β) with C2-N₃ donors under Yu’s gold-catalyzed glycosylation conditions.^5c The Boons group reported direct β-mannosylation with mannosamine derivatives equipped with 4,6-O-pyruvyl ketal albeit with very poor diastereoselectivity.^5d Demchenko and co-workers recently accomplished very high β-diastereoselectivity through remote 3-O-picolyl participation (by HAD) on 2-azido-2-deoxy mannopyranosyl donor.^2e In this ccontext, we set out to investigate our silver-assisted gold-catalyzed activation of ethynylcyclohexyl carbonate donors since the donors⁶ have a large reaction temperature range (-78→25 °C),^6b,6c use only catalytic activation (8 mol% each of chloro[tris(2,4-di-tert-butylphenyl)phosphite]gold (3) and AgOTf), and have broad substrate tolerance.^6d Accordingly, the C2-azido group was envisioned as a good functional group as it is easily accessible and can be converted to the -NHAc post-glycan synthesis. Thus, we have initially synthesized two glycosyl donors (1a,1b) and considered model aglycon 2a for our exploration of concentration dependent diastereoselective synthesis of β-glycosides.

Scheme 1.

Naturally occurring 1,2-cis mannosamine glycosides (A) and some literature methods to synthesize 1,2-cis mannopyranosides (B)

Accordingly, glycosidations were conducted between the glycosyl donor 1a and aglycon 2a at 50 mM and 10 mM concentrations to afford the disaccharide 4a with full diastereoselectivity towards α-mannoside (Entry 1, Table 1).^2e The side product 5^6a formed in the glycosidation could be easily separated from the reaction mixture by simple silica gel column chromatography. Similar reaction performed between the picoloyl donor 1b and aglycon 2a produced inseparable mixture of α,β-disaccharide 4b (1:2) at 50 mM concentration. We did not pursue the donor 1b as it is non-catalytic. However, the diastereoselectivity increased to 1:6 at 50 mM and 1:10 at 10 mM when the glycosylation was conducted with the donor containing C3-OBn moiety (6a) at 25 °C with no improvement even at low temperatures; increased to 1:10 at 10 mM and remained almost the same to afford the disaccharide 7a.⁷ The gold-catalyzed glycosidation at −25 °C between the donor 6b and the aglycon 2a afforded 1:15 (α:β) ratio of disaccharides 7b at 10 mM concentration. The C3-ONap donor 6c also produced similar ratio (1:14) of disaccharides 7c at 10 mM concentration and −25 °C. The gold-catalyzed glycosylation between the C3-OTBDMS donor 8a and the aglycon 2a gave α:β ratio of 1:7 (50 mM), 1:11 (10 mM) at ambient temperature and improved to 1:9 (50 mM), 1:18 (10 mM) at 0 °C and the ratio of disaccharides 9a remained almost same when the temperature was lowered further to −25 °C (Entry 6, Table 1). In this context, it is important to note that the C3-OTBDMS group showed diminished β-selectivity on mannopyranosyl sulphoxide donors even at −78 °C which shows that dramatic influence of the donor chemistry and C2-azide on the β-selectivity.^2f Substantial loss in the β-selectivity was noticed with sterically demanding C3-OTBDPS donor 10 to afford disaccharides 11 (1:3) (Entry 7, Table 1). Thus, optimum conditions were found to be 8 mol% each of Au-phosphite catalyst (3), AgOTf, 10 mM CH₂Cl₂ solution at 0 °C in the presence of 4Å molecular sieves powder. The stereochemical consequence of the glycosylation was noticed to be dependent on the temperature and the concentration of the reaction. The practicality of this highly diastereoselective synthesis of 2-azido-2-deoxy-mannosides was investigated with diverse acceptors (Scheme 2).⁷

Table 1:

Synthesis of 1,2-trans (α-) and 1,2-cis (β-) disaccharides¹

Entry	Donor	Disaccharide	α:β ratio at 25 °C		α:β ratio at 0 °C		α:β ratio at −25 °C
			50 mM	10 mM	50 mM	10 mM	50 mM	10 mM
1.	1a	4a	1:0 (88)	1:0 (86)	--	--	--	--
2.	1b	4b 2	1:2 (80)	--	--	--	--	--
3.	6a	7a	1:5 (88)	1:10 (85)	1:6 (89)	1:12 (82)	1:5 (88)	1:11 (82)
4.	6b	7b	1:6 (90)	1:10 (88)	1:8 (89)	1:13 (84)	1:6 (89)	1:15 (80)
5.	6c	7c	1:6 (91)	1:8 (85)	1:7 (90)	1:14 (89)	1:5 (89)	1:14 (83)
6.	8a	9a	1:7 (91)	1:11 (86)	1:9 (89)	1:18 (88)	1:8 (89)	1:19 (82)
7.	10	11	1:3 (81)	--	--	--	--	--

¹Red colour indicates the leaving group and its extrusion product and blue colour indicates the aglycon, Numbers in the parenthesis show the percentage of yields;

²Reaction carried out with 1.0 equivalent each of [Au]- and [Ag]- catalysts.

Scheme 2.

The substrate scope of the Synthesis of 2-azido-2-deoxy-1,2-cis-mannosyl glycosides. Numbers in the parenthesis indicate percentage yield and α:β ratio of diastereomers. Blue colour indicates the aglycon

For example, the mannosyl azide 8a could be efficiently and selectively glycosylated to several aglycons (2b-2j) to afford disaccharides and glycoconjugates. For instance, the glycosylation between donor 8a and the furanosyl aglycon 2b underwent a smooth coupling to afford the allyl disaccharide 9b in very high yield and selectivity. Similarly, a very high selectivity of 1:14 (α:β) was noticed with methyl 2,3,4-tri-O-benzoyl glucopyranoside 2c as aglycon to afford the disaccharide 9c in 90% yield. The α:β ratio, however, was compromised in the case of the more sterically challenging aglycons 2d-2g while nevertheless affording disaccharides 9d-9f and amino acid mannoconjugate 9g with yields ranging from 78–91%. Subsequently, the protocol was successfully tested with benzyl alcohol (2h), menthol (2i), and cholesterol (2j) as aglycons to afford 2-azido-2-deoxy-mannopyranosides 9h (86%, 1:7), 9i (92%, 1:16) and 9j (91%, 1:13) (Scheme 2).

In an attempt to understand the reaction, we have carried out variable temperature (VT) NMR experiments, which were performed by activating glycosyl donor 8a with a stoichiometric amount of [tris(2,4-di-tert-butylphenyl)phosphite]gold trifluoromethanesulfonate at −78 °C in CD₂Cl₂ in the NMR tube.⁷ No reaction was observed immediately after addition of 1.1 equivalents of the precooled activated gold triflate catalyst solution to 0.08 M 8a in CD₂Cl₂ and on warming by 10 °C increments until −30 °C when we observed partial formation of an α-glycosyl triflate 12 and carbonate 5. The glycosyl triflate 12 was characterized by a ¹H NMR resonance at δ_H = 6.01 ppm and by a DEPT-135 NMR resonance at δ_C = 104.9 ppm corresponding to its H-1 and C-1 atoms, while carbonate 5 showed two characteristic ¹H NMR resonances at δ_H = 4.72 and 4.29 ppm corresponding to the vinylic protons, and by a DEPT-135 NMR signal at δ_C = 85.3 ppm corresponding to the vinylic methylene group. On further warming to −10 °C all residual glycosyl donor 8a was consumed in favor of the carbonate 5 and the glycosyl triflate 12, with the latter nevertheless reaching a maximum at −20 °C above which decomposition set in. Notably, at −20 °C a new singlet resonating at δ_H = 6.45 ppm was observed, which showed an HSQC correlation to the DEPT-135 NMR resonance at δ_C = 136.8 ppm, which we tentatively assign to the vinylidene gold-complex 13, and which, in presence of moisture, could undergo protodeauration to produce cyclic carbonate 5.⁷

Importantly, little to no change in the ratio of anomers of unconsumed donor 8a was observed during the course of these VT NMR experiments, suggesting that both were consumed at a similar rate. The overall results of the VT NMR experiments with glycosyl donor 8a are summarized in Scheme 3.

Scheme 3.

VT NMR investigation of the activation of donor 8a with stoichiometric premixed silver triflate and chloro[tris(2,4-di-^t-butylphenyl)phosphite]gold in CD₂Cl₂ at −78 °C with incremental warming showing a) the intermediates observed in the absence of cyclohexanol, and b) the contrast with activation in the presence of cyclohexanol.

Next, we conducted VT NMR experiments with 8a in the presence of cyclohexanol as acceptor. No reaction was observed on addition of the activated catalyst at −78 °C and until −30 °C when formation of the cyclohexyl glycosides 15α and 15β and cyclic carbonate 5 was observed, with full conversion occurring by −10 °C. The cyclohexyl glycosides 15α,β were characterized by two DEPT-135 NMR resonance at δ_C = 97.7 and 97.0 ppm corresponding to the anomeric carbons and were formed with an approximate ratio of α:β = 1 : 1.17 with the configurational assignments made by a combination of ¹H and coupled gHSQCad experiments (see Supporting Information). Importantly, when the activation was conducted in the presence of the acceptor, glycosyl triflate 12 was not observed, suggesting either direct conversion of the donor to the products on activation or immediate conversion of any intermediate triflate to the products at the relatively high temperatures needed for the activation.

Overall, the [Au]/[Ag]-catalyzed glycosylation method provides a general means for accessing mannosamino glycosides in a stereoselective fashion. We applied this finding to the synthesis of the fully protected trisaccharide 21 corresponding to the capsular polysaccharide of Streptococcus pneumoniae type 19F.⁸ We hypothesized that the trisaccharide 21 can be synthesized from donor 8a and two easily accessible building blocks 16 and 17 (Scheme 3).⁷ The journey commenced with the synthesis of glucopyranosyl donor 16 and the rhamnopyranosyl acceptor 17. The glucopyranosylation between donor 16 and aglycon 17 took place smoothly to produce the desired Glcp(1→2)Rhap derivative 18 in 85% yield with α-selectivity. Reductive opening of the benzylidene acetal of the disaccharide 18 using NaCNBH₃ and I₂ afforded the C-4 alcohol 19 that was subjected to the silver-assisted gold-catalyzed glycosidation at 10 mM concentration and 0 °C to afford the trisaccharide 20 in 92% yield with good diastereoselectivity. Further, the azide was transformed to the acetamido moiety by reduction of compound 20 with 1,3-dithiopropane and Et₃N, followed by acetylation using Ac₂O and pyridine to give the trisaccharide 21.⁷ The ¹³C NMR spectrum of the trisaccharide 21 depicted the three anomeric carbons at δ 100.0, 97.5, 96.9 ppm and the J_C1’,H1’ coupling constant of 164 Hz confirming the β-mannopyranosyl configuration, while that of 172 Hz confirmed the presence of the α-glucopyranosyl residue. Further, the HRMS spectrum of trisaccharide 21 displayed a molecular ion with m/z of 1310.5875 [m/z Calcd. for C₇₅H₈₉NO₁₆SiNa = 1310.5848] consistent with 21 and confirming its synthesis.⁷

In summary, the present study underscores the successful identification of a silver-assisted gold-catalyzed activation of ethynylcyclohexyl 2-azido-2-deoxy-mannopyranosyl carbonates for the synthesis of β-mannosamine derivatives. The high β-selectivity was influenced by the temperature and the concentration of the reaction. The VT-NMR studies confirmed the presence of anomeric triflate from −30 to −10 °C. We have demonstrated utility of the procedure by synthesizing various β-mannosides and a capsular protected trisaccharide motif of Streptococcus pneumoniae type 19F in an efficient manner. Further application of the methodology for the synthesis of other complex β-ManNAc polysaccharides is currently underway.

Supplementary Material

General methods, Experimental procedures, characterization data and spectral charts of all compounds.

The Supporting Information is available free of charge on https://www.acs.org

Supporting_information.pdf

Scheme 4.

Synthesis of trisaccharide fragment of CPS of S. pneumonie

Data Availability Statement

The data underlying this study are available in the pub-lished article and its Supporting Information