Scope and limitations of 2-deoxy- and 2,6-dideoxyglycosyl bromides as donors for the synthesis of β-2-deoxy- and β-2,6-dideoxyglycosides.

It is shown that 2-deoxy- and 2,6-dideoxyglycosyl bromides can be prepared in high yield (72-94%) and engaged in glycosylation reactions with β:α selectivities ≥6:1. Yields of product are 44-90%. Fully armed 2-deoxyglycoside donors are viable, while 2,6-dideoxyglycosides require one electron-withdrawing substituent for high efficiency and β-selectivity. Equatorial C-3 ester protecting groups decrease β-selectivity, and donors bearing an axial C-3 substituent are not suitable. The method is compatible with azide-containing donors and acid-sensitive functional groups.

The difficulties associated with the syntheses of β-2-deoxy and β-2,6-dideoxyglycosides are related to the electron-rich nature of these donors, which favor α-selective, cationic pathways, and the absence of a C-2 substituent to bias β-selectivity.[1] In nature, glycosyltransferases convert α-linked donors to β-linked products by invertive displacement,[2] but few direct synthetic methods that employ aliphatic, oxygen-based acceptors have been reported.[3,4] The most intensively developed strategy involves introduction of a removable equatorial C-2 substituent to control stereoselectivity (eq 1).[5] While this strategy reliably provides high β-selectivities, the synthesis of the diastereomerically pure C-2-substituted donor and removal of the directing group increase the number of steps associated with the glycoside construction. Alternative solutions to this problem include stereoselective addition to glycosyloxiranes,[6] strategic use of hydrogen-bonding[7] and conformational biasing elements,[8] stereoselective palladium-catalyzed allylic alkylation,[9] rhenium-catalyzed glucal hydroalkylation,[10] anomeric O-alkylation,[11] and reduction of anomeric radicals.[12]

Herein we show that glycosyl bromides, which form the basis of the Koenigs–Knorr method[13] and are among the first activated donors known,[14] can be employed in the direct construction of β-linked 2-deoxy- and 2,6-dideoxyglycosides and should be considered as a potential solution to this problem. Glycosyl halides exist predominantly, or exclusively, in the α-configuration, and in a seminal advance, Paulsen and co-workers reported that heterogeneous silver salts could mediate the formation of β-glycosides from these α-linked donors.[15] The stereochemical outcome of this process is commonly rationalized by invoking silver-assisted S2 displacement of the halogen by the acceptor, and sequestration of the liberated halide ion, which prevents donor anomerization. This strategy has been employed to construct stereochemically challenging products, such as β-mannosides.[15b] However, its application toward the syntheses of β-2-deoxy- and 2,6-dideoxyglycosides is limited to five isolated examples,[16] and a systematic study of predictive value has not been reported, to our knowledge. The limited application of deoxyglycosyl halides may be due to difficulties associated with their synthesis and handling: these electron-rich donors[17] readily ionize, leading to hydrolysis, elimination, and other decomposition products. We show below that under carefully controlled conditions a broad range of 2-deoxy- and 2,6-dideoxyglycosyl bromides can be prepared in excellent yield and purity and employed in highly β-selective glycosylation reactions. Donor substituent effects on efficiency and stereoselectivity are delineated.

As glycosyl bromides are intermediate in reactivity between glycosyl chlorides and iodides, we focused on these donors. Classically, glucosyl bromides have been prepared by treatment of anomeric acetates with hydrobromic acid in acetic acid,[14] conditions that are almost certain to be incompatible with deoxy donors. Thiem[18] and Gillard[19] independently described the conversion of glycosyl acetates (including 2-deoxyglycosyl acetates)[18] to glycosyl bromides by treatment with trimethylsilyl bromide in chloroform or benzene. These conditions are attractive because the side product in the reaction, trimethylsilylacetate, is easily removed in vacuo (bp = 107 °C at 760 Torr), and aqueous workup of the glycosyl bromide product can be avoided. Accordingly, we investigated the transformation of the deoxyglycosyl acetates 1a–d to the bromides 2a–d by treatment with trimethylsilyl bromide (2.25 equiv) in dichloromethane at 0 °C (Table 1). After stirring for 20 min–2.5 h at 0 °C, the product mixture was concentrated in vacuo (100 mTorr, 0 °C, 10 min). Residual trimethylsilyl bromide and trimethylsilyl acetate were removed by codistillation with benzene under reduced pressure. Under these conditions, the glycosyl bromides 2a–d were reproducibly formed in yields of 72–94%, as determined by 1H NMR analysis against an internal standard. Solutions of 2a–2d in benzene-d6 at 24 °C were monitored by 1H NMR spectroscopy to assess their stability (Table 1, bottom). These studies revealed the stability ranking 2b > 2c ≈ 2d > 2a. The minor (≤8%) extent of decomposition of 2b–2d after 12 h suggested they may be viable in the glycosylation step. Attempted aqueous workup of the bromide 2d (dilution with saturated aqueous sodium bicarbonate solution) led to hydrolysis and decomposition. Chlorodeacetoxylation of the donor 1a (chlorotrimethylsilane, 5 equiv, 24 °C) did not proceed to completion. The donors 1e and 1f, which contain an axial C-3 substituent, did not form stable glycosyl bromides (Figure 1).

Table 1

Syntheses of Glycosyl Bromide Donors 2a–d

Determined by 1H NMR spectroscopy using mesitylene as an internal standard.

Figure 1

Donors that did not form stable glycosyl bromides.

We then sought to evaluate the suitability of the donors 2a–2d in glycosylation reactions and to elucidate the influence of substituents and protecting groups on efficiency and stereoselectivity (Table 2). We employed (−)-menthol (3a) as the acceptor. The donors 1a–1d were transformed to the corresponding glycosyl bromides 2a–2d, as described above, and then redissolved in dichloromethane. The resulting solution was added to a mixture of 3a (1.5 equiv), silver silicate (1.6 g/mmol donor), and activated 4 Å MS (860 mg/mmol donor) in dichloromethane at −60 °C. Under these conditions, the products 4a–4d were obtained in 53–81% yield (yields are based on the glycosyl acetates 1a–1d). The experiments outlined in Table 2 define the structural parameters of the donor that lead to high efficiency and β-selectivity. The fully armed 2,6-dideoxyglycoside donor 1a provided modest β-selectivity (β:α = 3.4:1), and the glycosylation proceeded in only 53% yield (entry 1). The modest yield and selectivity may be due to the lower yield in the formation of 2a (vide supra) and competitive ionization, which leads to α-selective pathways and decomposition. The partially deactivated donor 1b, which contains a C-4 benzoyloxy substituent, provided an 81% yield of product and 18:1 β-selectivity (entry 2). Notably, the isomeric donor 1c, which contains a C-3 benzoyloxy substituent, was selective for the α product (β:α = 1:1.5), and the glycoside 4c was isolated in 67% yield as a mixture of anomers (entry 3). The lower β-selectivity with the C-3-benzoyloxy donor 1c may be due to anchimeric assistance from the ester, which leads to a bridged oxacarbenium ion such as 6 and α-selective addition (eq 2).[20] The fully armed 2-deoxydonor 1d was a suitable substrate for the glycosylation and provided the product 4d in 74% yield and with 22:1 selectivity in favor of the β-anomer (entry 4). To probe the influence of temperature on selectivity and yield, this experiment was also conducted at −40 °C. Under these conditions, the product 4d was obtained in 77% yield and the selectivity in the reaction was only slightly diminished (β:α = 14:1, entry 5).

Table 2

Evaluation of Carbohydrate Substituents on Efficiency and Selectivity

Determined by 1H NMR or LC/MS analysis of the unpurified product mixture; see Supporting Information.

Isolated yield after purification by flash-column chromatography. Yield based on 1a–d.

Isolated as a mixture of α- and β-anomers.

Glycosylation conducted at −40 °C.

The scope of the glycosylation was then evaluated (Table 3). For all entries, 1.5 equiv of the acceptor was employed, and the yields are based on the glycosyl acetates 1b, 1d, and 1g. The α and β diastereomers were readily distinguished by the chemical shift and vicinal coupling constants of the anomeric proton.[21] Coupling of the donor 1d with 1,4-dioxaspiro[4.5]decan-8-ol (3b) provided the glycoside 4e in 90% yield and with 27:1 β-selectivity (entry 1). Coupling of N-Boc-l-serine methyl ester (3c) with 1d provided the glycoside 4f as a 6:1 mixture of β and α isomers in 80% yield (entry 2). This result is important because β-hydroxy carbonyl compounds are known to be poor acceptors in glycosylation reactions.[5m] The glucals 3d and 3e were coupled to the donor 1d in 44% and 46% yield, respectively, and with high β-selectivities (entries 3 and 4). The methyl glycoside 3f was glycosylated with the donor 1d in 78% yield and with 47:1 β-selectivity (entry 5). The azide-containing disaccharide 4j was obtained in 72% yield and with 13:1 β-selectivity from the azidosugar 1g and the acceptor 3f (entry 5). Finally, the donor 1b could be coupled with 1,4-dioxaspiro[4.5]decan-8-ol (3b), to provide the glycoside 4k in 90% yield and with 20:1 β-selectivity. The result in entry 6 demonstrates that the method is compatible with azide-containing donors. The results in entries 1–4 and 7 establish the method as compatible with acid-sensitive functional groups. It is noteworthy that hindered secondary alcohols, such as 3a, 3d, and 3e, are suitable acceptors in this transformation.

Table 3

Scope of the Glycosylation Reaction

Determined by 1H NMR or LC/MS analysis of the unpurified product mixture, see Supporting Information.

Isolated yield after purification by flash-column chromatography. Yield based on the glycosyl acetates 1b, 1d, and 1g.

Isolated as a mixture of anomers.

Structure of 1g:

Several experimental parameters are critical to the success of these reactions. First, it is essential that excess trimethylsilyl bromide is removed under strictly anhydrous conditions. Trace amounts of hydrobromic acid, formed by reagent hydrolysis, decompose the donors. Second, the bromide donors we have prepared are unstable toward TLC analysis; 1H NMR analysis provides a reliable method to assess conversion in the bromination step. The rates of bromination are dependent on the electron-density of the donor (as expected), and for all donors we investigated, except 1g,[23] this step was complete within 2.5 h at 0 °C. Finally, all of the trimethylsilylacetate byproduct must be removed prior to the glycosylation step to prevent regeneration of the glycosyl acetate. This is readily assessed by 1H NMR analysis of the unpurified glycosyl bromide (chemical shifts for trimethylsilylacetate in benzene-d6: δ 1.71, 0.23).

In summary, we have demonstrated that 2-deoxy- and 2,6-dideoxyglycosyl bromides can be employed in high-yielding, β-selective glycosylation reactions and delineate several useful reactivity trends. Fully armed 2-deoxyglycoside donors are suitable, while 2,6-dideoxyglycosides require at least one electron-withdrawing substituent. Equatorial C-3 ester protecting groups decrease β-selectivity,[22] and donors bearing axial C-3 heteroatom substituents do not form stable activated bromides. Hindered and deactivated acceptors, acid-sensitive functional groups, and azide-containing donors are compatible with the method. The use of glycosyl acetates is strategically advantageous, as it provides protection of the C-1 hydroxyl group in the event that additional manipulations of the donor are required. The studies outlined herein should motivate the application of glycosyl bromides in this difficult bond construction.