Highly Stereoselective Synthesis of 2-Azido-2-Deoxyglycosides via Gold-Catalyzed S_N2 Glycosylation

Abstract

Highly stereoselective synthesis of 2-azido-2-deoxyglucosides and 2-azido-2-deoxygalactosides is achieved via a gold-catalyzed S_N2 glycosylation. The glycosyl donors feature a designed 1-naphthoate leaving group containing an amide group. Upon gold activation of the leaving group, the amide group is optimally positioned to direct an S_N2 attack by an acceptor via H-bonding interaction. Both 2-azido-2-deoxyglucosyl/galactosyl donor anomers can undergo stereoinversion at the anomeric position, affording the opposite anomeric glycoside products with excellent levels of stereoselectivity or stereospecificity and in mostly excellent yields. This S_N2 glycosylation accommodates a broad range of acceptors. The utility of this chemistry is demonstrated in the synthesis of a trisaccharide featuring three 1,2-cis-2-azido-2-deoxyglycosidic linkages.

Graphical Abstract

Introduction

2-Amido/amino-2-deoxy-d-glycopyranosyl units are found in many biologically important poly-/oligasaccharides, glycoconjugates, and glyco-based antibiotics and medicines (see Scheme 1a for examples).^1–5 Their efficient assembly with high levels of anomeric stereoselectivity^6–11 is essential to glycoscience research and the development of glycomedicine. 1,2-trans-2-amido/amino-2-deoxyglycosides can be reliably synthesized through the neighboring group participation of a C(2)-amido/imido/alkoxycarbonylamino group, albeit with drawbacks (Scheme 1b).^10,12–14 In comparison, the stereoselective construction of 1,2-cis-counterparts remains challenging.¹⁵ In the 1970s, Paulsen et al.¹⁶ and Lemieux and Ratcliffe¹⁷ pioneered the use of 2-azido-2-deoxyglycosyl donors for stereoselective 1,2-cis glycosylation. Despite the 2-azido group serving as an ideal masked amino group due to its relative stability, nonparticipating property, and ease of transformation to the amino/acetamido group, the glycosylation reactions exhibit mostly moderate to poor 1,2-cis selectivity^18–20 and low/moderate yields.^21,22 Several innovative solutions have since been documented, including the use of ring-fused 2,3-oxazolidinone thioglycoside donors,^23–25 the Lewis/Brønsted acid-catalyzed α-glycosylation of benzylidene-protected 2-aminoglycosyl trihaloacetimidate donors,^26–29 and the employment of steric shielding or remote participation,³⁰ but they can be limited in scope,^30,31 afford moderate yields²⁵ due to side reactions,^31,32 and/or exhibit moderate/varied α/β selectivity. As such, considering the biological significance of aminoglycosides, there is still a great need for developing high-yielding and highly stereoselective synthesis of 1,2-cis-2-amido/amino-2-deoxyglycosides that applies to a broad range of acceptors.

Scheme 1 |.

Background of this work.

In 2021, we published a study introducing an S_N2 glycosylation strategy that employs a directing-group-on-leaving-group strategy (DGLG)^33,34 based on Yu’s glycosyl ortho-alkynylbenzoate system.^35,36 Recently, we reported that a second-gen approach improved upon the original work by replacing the oxazole-functionalized O-alkynylbenzoate with a synthetically much more accessible amide-functionalized 1-naphthoate in the glycosyl donor, which exhibited enhanced S_N2 characteristics in the construction of both 1,2-cis and 1,2-trans glycosidic linkages as well as both 2-deoxyglucoside anomers.^37,38 Based on these results, we speculated that this DGLG strategy could effectively address the long-standing challenges of achieving high stereoselective glycosylation with 2-azido-2-deoxyglycosyl donors. In this work, we disclose the synthesis of 2-azido-2-deoxyglucosides and 2-azido-2-deoxygalactosides based on this consideration. The glycosylation method offers several desirable features, including mild reaction conditions, accommodation of a wide range of acceptors, high yielding, and high levels of S_N2 characteristics (Scheme 1c). Both 1–2-cis- and 1,2-trans-2-azido-2-deoxyglycosides can be synthesized with high efficiency and high to excellent stereoselectivity.

Experimental Section

General procedure for gold-catalyzed S_N2 glycosylation

A donor (0.06 mmol), an acceptor (2.0 equiv, 0.12 mmol), and 100 mg grounded white drierite were added sequentially to a vial. Trifluorotoluene (PhCF₃, 0.45 mL) was then injected. The vial was capped with an open-top screw cap with a silicon liner, sealed with parafilm, shaken, and cooled to the corresponding temperature. Meanwhile, a PhCF₃ solution (0.3 mL) of (PhO)₃PAuCl (0.1 equiv, 0.006 mmol, 3.26 mg), AgNTf₂ (0.1 equiv, 0.006 mmol, 2.33 mg), and HNTf₂ (0.2 equiv, 0.012 mmol, 3.37 mg) was injected into the vial. The reaction progress was monitored by thin-layer chromatography. Upon completion, a dichloromethane (DCM; 0.3 mL) solution of tetrabutylammonium chloride (1.0 equiv, 0.06 mmol, 16.7 mg) and N,N-diisopropylethylamine (1.0 equiv, 0.06 mmol, 7.8 mg) was used to quench the reaction, and the reaction mixture was filtered through a Celite pad before column chromatography.

Results and Discussion

The desired 2-azido-2-deoxyglycosyl 1-naphthoate donors of type 1 (e.g., 1a in Table 1) were readily prepared in two steps—alkylation and Sonogashira coupling—from the corresponding 2-azido-2-deoxyglycosyl halides^39–41 and commercially available 8-bromo-1-naphthoic acid (Scheme 2, see the Supporting Information for details). The two-step yields range from 70% to 85%.

Table 1 |.

Conditions Optimizations^a

Entry	Acceptor	Ligand	Temp./time	Yieldb (a/β)c
1	1.2 equiv A1	IMes	0 °C/3 h	75% (19/1)
2	1.2 equiv A1	IMes	−20 °C/24 h	78% (>30/1)
3	1.2 equiv A1	Ph3P	−20 °C/12 h	80% (>30/1)
4	1.2 equiv A1	P(OPh)3	−20 °C/8 h	84% (>30/1)
5	2.0 equiv A1	P(OPh)3	−20 °C/5 h	94% (>30/1)
6d	2.0 equiv A1	P(OPh)3	−20 °C/24 h	76% (>30/1)
7e	2.0 equiv A1	P(OPh)3	−20 °C/24 h	74% (>30/1)
8	1.2 equiv A2	IMes	0 °C/5 h	70% (α only)
9	1.2 equiv A2	IMes	−20 °C/24 h	72% (α only)
10	1.2 equiv A2	Ph3P	−20 °C/12 h	70% (α only)
11	1.2 equiv A2	P(OPh)3	−20 °C 9 h	77% (α only)
12	2.0 equiv A2	P(OPh)3	−20 °C/6 h	85% (α only)

^aReaction was stirred in a cooling bath before being quenched by 1.2 equiv ⁿBu₄NCl and N,N-diisopropylethylamine.

^bThe yield was based on the donor.

^cAnomeric ratio was determined by ¹H NMR analysis of the crude reaction mixture.

^dWithout HNTf₂.

^eDCM (0.08 M) as the solvent.

Scheme 2 |.

Synthesis of 2-azido-2-deoxyglycosyl 1-naphthoate donors.

We initiated the conditions study by examining the reaction between 2-azido-2-deoxy-d-glucosyl 1-naphthoate donor 1a and the methyl α-d-glycopyranoside A1—a primary acceptor. First, the reaction was conducted using 1.2 equiv of A1 in the presence of 10 mol % IMesAuCl/AgNTf₂ and 20 mol % HNTf₂, in PhCF₃ (0.08 M) at 0 °C. Within 3 h, the reaction was complete, yielding methyl (3,4,6-tri-O-benzyl-2-azido-2-deoxy-α-d-glucopyranosyl)-(1→6)-2,3,4-tri-O-benzyl-α-d-glucopyranoside 2a in 75% nuclear magnetic resonance (NMR) yield, with the con-comitant generation of the byproduct 3.³⁷ The reaction showed a favorable α/β ratio of 19/1 (Table 1, entry 1). Lowering the reaction temperature to −20 °C and prolonging the reaction time resulted in an improved α/β ratio of >30/1 (Table 1, entry 2). More acidic gold catalysts such as Ph₃PAu⁺ and (PhO)₃PAu⁺ improved the yield slightly and sped up the reaction, notably in the latter case (Table 1, entries 3 and 4). By increasing the amount of A1 to two equivalents, the reaction yield was substantially improved to 94% while maintaining the excellent α selectivity (entry 5). In the absence of 20% HNTf₂, the reaction proceeded much more slowly, requiring 24 h to reach complete consumption of 1a, and the yield was a lower 76% (entry 6). This outcome is consistent with our previous conclusion³⁷ as well as that of Zhu and Yu,⁴² that is, a strong acid additive facilitates protodeauration and hence catalyst regeneration, there-by speeding up the reaction. A similarly sluggish and lower-yielding reaction was observed when the reaction solvent was DCM (entry 7). This solvent effect was previously observed³⁷ and rationalized to be due to the much higher solubility of the interfering byproduct 3 in DCM than in PhCF₃. For the more hindered and hence more challenging secondary acceptor A2, similar trends were observed in the stereoselective formation of the diglucoside product 2b under these varied conditions (entries 8–12). Moreover, the reactions exhibited complete α-selectivity, and by using two equivalents of the acceptor, the yield was 85% (entry 12).

With the optimized conditions in Table 1, entries 5 and 12 in hand, we further examined the reaction of donor 1a with a range of additional acceptors. As shown in Table 2, remarkably, all the cases, i.e., 2c–2l exhibited exclusive α selectivities. For these cases employing primary acceptors (2f, 2g, and 2k), the reaction yields are above 90%, as is the case in 2a. For the cases using challenging secondary acceptors (2d, 2e, 2i, and 2l), the reaction yields (71–85%) are comparatively lower. On the other hand, both the yields of 2h and 2j were 94%. With tert-butanol as the acceptor, the product was obtained in 84% yield (2c). The acceptors included nonsugar alcohols tert-butanol (2c), l-menthol (2d) and cholesterol (2e), primary glycosyl acceptors derived from galactose (2f), glucose (2g), and arabinose (2k), and secondary acceptors derived from glucose (2h, 2i), rhamnose (2j), and xylose (2l). Of note is that the benzoyl-protecting groups in the cases of 2g and 2h did not affect the outstanding reaction outcomes, suggesting that these acceptors are sufficiently nucleophilic despite inductive deactivation.

Table 2 |.

Reaction Scope with 2-Azido-2-Deoxy-β-d-Glucopyranosyl Donors. Reaction Conditions: Donor 1 (0.06 mmol), Acceptor (0.12 mmol), (PhO)₃PAuCl/AgNTf₂ (0.006 mmol), HNTf₂ (0.012 mmol), PhCF₃ (0.08 M), Drierite, −20 °C

We also prepared the 2-azido-2-deoxy donor variants of 1a with the benzyl groups at 6-O and 4-O positions replaced by acetyl groups, respectively. As shown in the cases of 2m–2q, the glycosylation reactions again exhibited exclusive α-selectivities with both primary and secondary glycoside acceptors, and the reaction yields ranged from 76% to 94%.

We then applied our methodology to the synthesis of 1,2-cis-2-azido-2-deoxygalactosides. However, the reactions of per-benzylated 2-azido-2-deoxy-β-d-galactopyranosyl donors 1b were surprisingly slow at −20 °C. Elevating the reaction temperature to 0 °C led to a complete reaction in 3–24 h. As shown in Table 3a, the glycosylation reactions proceeded with mostly excellent levels of stereoinversion, affording 1,2-cis-2-azido-2-deoxygalactosides 2r, 2s, and 2u–2y exclusively or with >20:1 α/β selectivities and in mostly excellent yields. The exception was the case of 2t with an α/β ratio of 12/1. Similar to the 2-azido-2-deoxyglucoside cases, a variety of acceptors including secondary (2u–2w and 2y) and electronically deactivated (2t and 2v) were readily tolerated. However, this strategy is not suitable for the synthesis of β−2-azido-2-deoxymannosides.

Table 3 |.

Reaction Scope with Additional Donors. Reaction Conditions: Donor 1 (0.06 mmol), Acceptor (0.12 mmol), (PhO)₃PAuCl/AgNTf₂ (0.006 mmol), HNTf₂ (0.012 mmol), PhCF₃ (0.08 M), Drierite, 0 °C.

^aStarting from the donor with a 15/1 α/β ratio.

^b−20 °C.

^cStarting from the donor with a 22/1 α/β ratio.

In addition to these conversions of β-donors to 1,2-cis-α−2-azido-2-deoxyglycosides, we wondered whether this strategy would permit the stereoselective synthesis of 1,2-trans-2-azido-2-deoxyglycosides^43–45 by using the corresponding 1,2-cis-2-azido-2-deoxy donors. Despite the synthesis of 1,2-trans-2-amido-2-deoxyglycosides, as discussed previously, can be achieved reliably via neighboring group participation, the multiple steps needed for converting a suitable participating group to the often-desired acetamido and the harsh reaction conditions employed therein often lead to diminished synthetic efficiency. The stereoselective synthesis of 1,2-trans-2-azido-2-deoxyglycosides via this S_N2 glycosylation strategy represents a valuable and unified approach to stereoselective access to both anomers of 2-azido-2-deoxyglycosides. As shown in Table 3b, the 2-azido-2-deoxymannoside 2z was formed with exclusive stereoinversion and in 78% yield. The α-counterpart of glucosyl donor 1a was prepared with an α/β ratio of 15/1. As shown in Table 3c, its reaction with the primary acceptor A1 resulted in near inversion of the α/β ratio of the product 2aa at 1:16, suggesting high levels of S_N2 glycosylation. With the challenging A2 as the acceptor, the α/β ratio of the product 2ab is 1/8, which amounts to an estimated α/β ratio of 1:18 if the pure α-donor is used. In addition, its reaction with an arabinose-derived acceptor afforded the product 2ac in 95% yield and with an α/β ratio of 1/12. In the cases of the α-galactosyl donor (α/β = 22/1, Table 3d), the reactions with primary acceptors were also highly stereoinvertive, affording 2ad–2af in excellent yields and with ≥20:1 β/α selectivities.

To demonstrate the synthetic utility of this chemistry, we set out to prepare the synthetically challenging trisazidoglucoside 6 containing three 1,2-cis-2-azido-2-deoxy glycosidic linkages. The pseudoiterative synthesis sequence is shown in Scheme 3: the first such glycosidic linkage in 4 was formed with exclusive α-selectivity and in 92% overall yield upon coupling the 6-O-acetyl donor 1c with pent-4-en-1-ol (2.0 equiv), followed by basic methanolysis. Subsequent glycosylations using the 4-O-acetyl donor 1d and the per-O-benzylated donor 1a, respectively, again exhibited complete anomeric configuration inversion and high reaction efficiency. In both cases, two equivalents of the acceptor (i.e., 4 or 5) were employed. After the reaction, the excess acceptor in the case of 5 was recovered.

Scheme 3 |.

Synthesis of the trisazidoaccharide 6. Two equivalents of acceptors were used, and the yields were calculated based on the donor. ^a1.0 equiv 5 was recovered after the reaction.

The high level of stereoinversions observed in most of the cases, that is, the conversion of β donors to α glycosides and vice versa, is consistent with the directed S_N2 glycosylation mechanism. As shown in Scheme 4a, the directed S_N2 reaction of the cyclized and hence activated intermediate β-A-Au generated from a β-donor (i.e., β−1) should afford the stereo-inverted 2-azido-2-deoxyglycoside product α−2. A minor S_N1 pathway leading to the oxocarbenium intermediate B may occur, which is trapped by the stoichiometric and more nucleophilic amide moiety from species bearing this functional group to form the trapped intermediate α-C-Au/α-C. Intermediate C favors the α configuration due to the anomeric effect. In addition, α-C-Au can be generated directly from β-A-Au via intramolecular stereoinvertive amide attack. In the presence of a good nucleophile, for example, a primary alcohol donor, this intermediate undergoes further glycosylation to deliver β−2. In contrast, with a poor nucleophilic acceptor such as a secondary alcohol, this glycosylation does not occur. Consequently, when using β donors, employing a primary alcohol acceptor results in compromised α/β selectivity but increased yield. When reacting with secondary alcohols, complete stereo inversion is maintained, but the yield can be compromised. In the case of 2t, the α/β ratio of 12/1 is much lower than that of 2r. We attributed the moderate stereoselectivity to the decreased nucleophilicity of the tri-O-benzoyl-protected acceptor, which leads to increased formation of α-C-Au/α-C. The acceptor, despite being inductively deactivated, is a primary alcohol and hence sterically viable to attack α-C-Au/α-C. As such, a significant amount of β−2t is formed.

Scheme 4 |.

Proposed mechanism rationalizing the observed selectivity.

We noticed that the β/α ratio of 2aa (16/1) is higher than that of 2ab. The former is formed from an α-glucosyl naphthoate donor and the primary acceptor A1, and the latter is formed from the same donor and the secondary acceptor A2. These results are contrary to the trend observed using β-donors and not consistent with the rationale put forth in Scheme 4a but can be rationalized in Scheme 4b. In this case, β-C-Au was generated from the cyclized intermediate α-C-Au via intramolecular stereoinvertive amide attack and should exhibit increased electrophilicity compared to its α-counterpart. As a result, it can glycosylate secondary acceptors. In comparison to using A1 as the acceptor, the glycosylation of A2 led to the increased formation of β-C-Au due to a comparably slow S_N2 process. As such, more α−2 can be formed via β-C-Au, resulting in a lower β/α ratio in the case of 2ab.

Conclusion

In summary, a gold-catalyzed S_N2 glycosylation method was achieved for highly stereoselective synthesis of 2-azido-2-deoxyglucosides and 2-azido-2-deoxygalactosides. This method employs a 1-naphthoate leaving group featuring an amide directing group. Upon gold activation of the leaving group, the amide is optimally positioned to direct an S_N2 attack by an acceptor via H-bonding interaction. From 1,2-trans-2-azido-2-deoxyglycosyl donors, this chemistry provides highly stereoselective and efficient construction of 1,2-cis-2-azido-2-deoxyglycosidic linkages. In most cases, exclusive 1,2-cis stereochemistry is achieved. This approach is also applicable to the stereoselective construction of 1,2-trans-2-azido-2-deoxyglycosides. A broad range of acceptors is allowed. The utility of this chemistry was demonstrated in the synthesis of a trisaccharide featuring three 1,2-cis-2-azido-2-deoxyglycosidic linkages.