Stereospecific Synthesis of 2-Deoxy-2-fluoroglycosides via S_N2 Glycosylation

Abstract

A directing-group-on-leaving-group strategy (DGLG) was applied to stereospecific synthesis of 2-deoxy-2-fluoroglycosides via S_N2 glycosylation. This strategy utilized 2-deoxy-2-fluoroglycosyl donors featuring a chromenone-based leaving group, which is activated rapidly under exceedingly mild conditions to achieve complete stereochemical inversion at the donor anomeric position. This method readily applies to 2-deoxy-2-fluoroglucose, 2-deoxy-2-fluorogalactose, 2,6-dideoxy-2-fluoroglucose, 2-deoxy-2-fluoroarabinose, and 2-deoxy-2-fluororibose, delivering high yields and complete stereochemical inversion, despite the presence of a neighboring deactivating fluoro group and the potential adverse fluorine directing effect. This method addresses the unmet challenges in constructing 1,2-cis glycosidic linkages in 2-deoxy-2-fluorosugars with stereospecificity. It also represents a major advance by enabling stereospecific synthesis of 2-deoxo-2-fluoroarabinofuranosides and 2-deoxy-2-fluororibofuranosides with either an α or a β anomeric configuration, showcasing the versatility of the DGLG approach. It offers by far the only stereospecific construction of these fluorinated glycosides of significant interest in glycoscience research. With the employment of excess donors permitted, this approach can be readily adopted to multistep sequential glycosylations, enabling streamlined and stereospecific access to complex fluorinated glycans.

Graphical Abstract

Introduction

Fluorinated sugars^1–4 have attracted considerable interest in medicinal chemistry (Scheme 1a)^5–7 and are also employed as positron emission tomography (PET) tracers⁸ and probes in chemical biology research.⁹ The replacement of a 2-OH group in a sugar ring with a fluorine in 2-deoxy-2-fluoroglucosides affords enhanced stability to the glycosidic bond, enhances anomeric effect, increase lipophilicity, alter binding affinity, and improves pharmacokinetics without significantly affecting sterics and H-bond accepting capacity.⁷ Various synthetic advances have been achieved in installing fluorine in monosaccharides,^1,10–13 but stereoselective glycosylation employing fluorinated donors and especially the 2-deoxy-2-fluoroglucosyl ones remains challenging and demands innovative solutions.

Scheme 1 |.

(a) Fluorinated sugars: prior stereoselective synthesis and our design.

The Gilmour group achieved fluorine-directed synthesis of 1,2-trans 2-deoxy-2-fluoropyranosides with moderate to excellent selectivities and mostly moderate yields (Scheme 1b1).^14–17 For the synthesis of more challenging 1,2-cis α−2-deoxy-2-fluoropyranosides, in 2021 the Nguyen lab reported a phenanthroline-catalyzed stereoselective synthesis (Scheme 1b2).¹⁸ Despite the fact that this method represents an important breakthrough, the α/β ratio in most cases was found to be no more than 13:1, and yields were mostly no more than 85%. Furthermore, the reaction takes 24 h, and 3 equiv of acceptors are used. On the other hand, stereoselective reactions of 2-deoxy-2-fluorofuranosyl donors are seldom studied. Among the three recently reported cases (Scheme 1c), one led to the highly selective formation of an α−2-deoxy-2-fluoroxyloside 1,¹⁹ but the other two afforded 2-deoxy-2-fluoroarabinoside 2 with β/α ratios of 5/1.^19,20 Of importance to note is that in all these prior efforts, excess acceptors (1.2–3 equiv) were employed to enhance yield and increase stereoselectivity. However, this practice is not desirable in sequential glycosylations, where acceptors become increasingly expensive. This is particularly true in automated glycan assembly²¹ based on solid-phase synthesis, where donors are often used in significant excess. To this end, the need to develop novel strategies for glycosylation of 2-deoxy-2-fluoro donors that are stereospecific, high-yielding, employing donors in excess, and broadly applicable remains unfulfilled. A breakthrough of this nature would, in turn, significantly facilitate access to fluorinated oligosaccharides either via sequential solution-phase or automated solid-phase synthesis and accelerate their applications in biomedical research.^22,23

For the past several years, our lab has pursued a directing-group-on-leaving-group (DGLG) strategy for the implementation of a general S_N2 glycosylation.^24–30 The first two approaches relied on gold catalysis,^31–36 but the reactions typically require 24 h due to impeded catalyst turnover, and excess acceptors are employed to improve the reaction rate and yield. Moreover, the scopes remain to be broadened. Our further implementation employs a stoichiometric approach.³⁷ As shown in Scheme 1d, the donor 3 features a chromenone-based leaving group, on which a diisopropylphosphinoyl moiety is installed judiciously as a directing group. With Y = OBn or H, 3 is activated upon a (Coll)₂Br⁺ NTf₂⁻-promoted electrophilic cyclization, affording the activated donor A. This activation is facile and complete in less than 10 min at −40 °C. The desired S_N2 glycosylation is realized upon a backend attack at the anomeric center by an acceptor. This attack is directed by the diisopropylphosphinoyl group via the formation of a strong H-bond³⁸ and facilitated by the proper positioning of the directing group in the leaving group via a rigid connection to the sugar ring (highlighted in blue in A). The glycosylation proceeds to completion in dichloromethane (DCM), in 1 h, and at −40 °C and exhibits excellent stereoselectivity in a mostly S_N2 fashion. These reaction characteristics are ideally suited for automated synthesis, and the reaction is successfully applied in the construction of challenging 1,2-cis glycosidic linkages in automated glycan assembly.^21,37 We surmise that this approach might be amenable to stereospecific synthesis of 2-deoxy-2-fluoropyranosides and 2-deoxy-2-fluorofuranosides. Such a method would enable the incorporation of 2-deoxy-2-fluoro sugars in expedient sequential solution-phase or automated solid-phase oligosaccharide synthesis and thus expand the scope of glycans accessible to nonspecialists for biomedical research. It is, however, not clear to us that replacing a 2-OBn(H) group in the chromen-1-one-based donor 3 (Y = OBn or H) with a more electron-withdrawing fluoro group could be tolerated or not. In addition, the F-directing effect Gilmour^14–16 observed could adversely impact the reaction stereoselectivity in the construction of 1,2-cis glycosidic linkages.

Results and Discussion

At the outset, we studied the reaction of the 2-deoxy-2-fluorogalactosyl donor 3a with methyl 2,3,4-tri-O-benzylglucoside (4) (Table 1). 3a was prepared by following the previously reported protocols by installing the polar directing group at the end,³⁷ which permits easy purification. The optimized conditions use 1.5 equiv of 3a, 1.8 equiv of (Coll)₂Br⁺ NTf₂, at −40 °C, in anhydrous DCM, and with 4 Å molecular sieves (MS). The reaction proceeded to completion in 1 h and was extremely clean, affording the desired 2F-Gal(α1→6)Glc product 5a in 99% yield and with exclusive α-configuration at the newly formed glycosidic bond (entry 1). This result compares favorably to the reported one (85% yield, α/β = 11:1, 24 h, 3 equiv of acceptor),¹⁸ and the reaction conditions—fast reaction (1 h), donor access, resin-suitable solvent (DCM)—should be conducive to automated glycan assembly.^21,37 Using lesser amounts of 3a (1.2 equiv) and (Coll)₂Br⁺ NTf₂ (1.3 equiv) worked equally well (entry 2). We opted to use higher amounts in the optimal conditions (entry 1) to enhance the tolerance of a broad range of donors/acceptors in achieving excellent yields and acceptor conversion. Increasing the reaction temperature to −20 °C (entry 3) or 0 °C led to decreased α/β selectivities, and solvents known to participate in glycosylation reactions^39,40 such as CH₃CN (entry 5) and Et₂O (entry 6) led to poor conversions, poor yields, and low stereoselectivity.

Table 1 |.

Reaction Conditions Optimization

Entry	Change from “Standard Conditions”	Resultsa
1	None	Full conversion; 99% yield (5a); α only;
2	1.2 equiv 3a and 1.2 equiv (Coll)2Br+NTf2− was used	Full conversion; 99% yield (5a); α only;
3	−20 °C, 1 h instead of −40 °C, 1 h	1.5% 4 left; 96% yield (5a); α/β = 58:1;
4	0 °C, 1 h instead of −40 °C, 1 h	4.5% 4 left; 93% yield (5a); α/β = 18:1;
5	CH3CN instead of DCM	67% 4 left; 17% yield (5a); α/β = 4.9:1;
6	Et2O instead of DCM	44% 4 left; 14% yield (5a); α/β = 3.5:1;

^aYield and anomeric ratio determined by ¹H NMR analysis of the crude reaction mixture. (Coll)₂Br⁺NTf₂⁻: bis(symcollidine)bromine(I) bis(trifluoromethanesulfonyl)imide.

To explore the reaction scope, 3a was reacted smoothly with five additional acceptors, delivering the 2-deoxy-2-fluoropyranosides 5c–5g in ≥90% yield and with complete anomer stereochemistry inversion either after 1 h at −40 °C or after 4 h at −60 °C (Table 2a). Notably, sterically hindered secondary alcohols were readily allowed. In the case of 5g, the acceptor sulfide moiety was compatible with the electrophilic Br reagent under the reaction conditions, and running the reaction at −60 °C resulted in better conversion of the acceptor. In the case of 5b, an acetyl group at the 6-O position replaced the Bn group in 3a and, although disarming, posed no issue.

Table 2 |.

The Scope of 2-Deoxy-2-fluoropyranosides Accessible via the DGLG Approach^a

^aIsolated yield reported. For reactions running at −60 °C, 2.0 equiv of donor and 2.4 equiv of (Coll)₂Br⁺ NTf₂⁻ were used.

^b8% of the acceptor remained.

This S_N2 glycosylation with a β−3,4,6-tris-O-benzyl-2-deoxy-2-fluoroglucosyl donor also worked smoothly, delivering the fluorinated diglucosides 5h–5l in yields ranging from 86% to 99% and with complete α-configuration (Table 2b). Both primary (5h–5i) and secondary acceptors (5j–5l) were tolerated. In comparison, 5j was synthesized previously in 68% yield and with α/β = 10:1.¹⁸ Extending the pyranosyl donor scope revealed that a β−2,6-dideoxy-2-fluoroglucosyl donor also underwent complete S_N2 glycosylation with both primary and secondary acceptors to deliver 5m–5s with an α-anomeric configuration exclusively and in 80%–99% yields (Table 2c).

Of note is that the α configurations of these products were established either by comparison to a literature report¹⁸ and/or by measuring their ¹J_C1-H1 values, which ranged from 170 to 176 Hz.⁴¹ Examples of crude ¹H nuclear magnetic resonance (NMR) spectra are provided in the Supporting Information to support the exclusive α-selectivity in these glycosylation reactions.

Our attempts to study the synthesis of the β-counterparts of the pyranosides in Table 2 were thwarted by the difficulty of accessing the requisite pure α-donors.

We then examined the scantily studied 2-deoxy-2-fluorofuranosides (Table 3). We first employed a 2-deoxy-2-fluoroarabinofuranosyl donor, that is, 7. Both the α- and β-anomers of this donor, obtained in pure form, were reacted with an identical set of six acceptors, and the results are shown in Table 3a,b, respectively. In comparison to the reported synthesis of the arabinofuranoside 2 (α/β ratio of 1/5, Scheme 1c),^19,20 β−2 was synthesized from α−7 in 99% yield and stereospecifically. Conversely, α−2 was synthesized from β−7 with exclusive stereoinversion and in nearly quantitative yield. The absence of the other anomer in these two reactions was evident from the crude ¹H NMR spectra (Figure 1) and highlighted the extraordinary stereocontrol rendered by our chemistry. The ¹H NMR spectra of β−2 and α−2 were identical to the reported major isomer and the minor isomer, respectively,^19,20 which supported our structural assignments. Moreover, our detailed spectroscopic analysis confirmed that the ¹J_C1-H1 values—175 Hz for β−2 and 176 Hz for α−2—were not useful for assigning anomeric configuration of furanosides, but the ³J_H1-F2 values—<1 Hz for β−2 possessing trans H1-F2 and 12.1 Hz for α−2 possessing cis H1-F2—are drastically different and might be employed to assign anomeric configurations of 2-deoxy-2-fluorofuranosides. The reaction of α−7 with five additional acceptors afforded additional 1,2-cis arabinofuranosides β−8a–β−8e with complete stereochemical inversion at the anomeric center and in mostly excellent yields (Table 3a). In the challenging cases employing secondary acceptors (β−8c–β−8e), the reactions proceeded smoothly at −60 °C. Not surprisingly, the reactions of β−7 were also highly efficient and stereospecific, affording the opposing 1,2-trans 2-deoxy-2-fluoroarabinofuranosides α−8a–α−8e in ≥91% yield and with complete stereochemical inversion (Table 3b). The configuration assignments of these anomeric pairs were based on the S_N2 nature of the glycosylation and substantiated by the consistently contrasting ³J_H1-F2 values, that is, <1 Hz for β−8 possessing trans H1-F2 and 11.4–12.8 Hz for α−8 possessing cis H1-F2.

Table 3 |.

The Scope of 2-Deoxy-2-fluorofuranosides Accessible via the DGLG Approach^a

^aReaction conditions: 1.5 equiv of donor, 1.0 equiv of acceptor, 1.8 equiv of (Coll)₂Br⁺ NTf₂⁻, 4 Å MS in anhydrous DCM, at −40 °C, and 1 h. For reactions at −60 °C, 2.0 equiv of donor and 2.4 equiv of (Coll)₂Br⁺ NTf₂⁻ were used, and the reaction time was 4 h. Isolated yield was reported.

Figure 1 |.

NMR evidence of the reaction stereospecificity.

Furthermore, α- or β−2-deoxy-2-fluororibosyl donor 9 also reacted smoothly with primary acceptors to deliver α−10a–α−10c (Table 3c) and β−10b and β−10c (Table 3d), respectively, in good to excellent yields and with complete anomeric configuration inversion. The reaction of β−9 with (−)-menthol, a secondary alcohol, was also successful, delivering α−10d in a nearly perfect outcome (Table 3c). However, with sugar-based secondary acceptors, the reactions experienced incomplete conversions and hence low yields (≤65%), despite exhibiting complete inversion of anomeric configurations. The anomeric configurations of these furanosides were established by a combination of NMR spectroscopy and the assumption of the S_N2 chemistry. We noted that α−10 possessing 1,2-trans H1-F2 exhibited substantially smaller ³J_H1-F2 values (3.1–5.5 Hz) than β−10 possessing 1,2-cis H1-F2 (10.6 and 11.0 Hz), offering further support to the expectation that this coupling constant can be used to assign the anomeric configurations of 2-deoxy-2-fluorofuranosides, with a value of >10 Hz likely pointing to a 1,2-cis H1-F2 relationship. This conjecture remains to be verified with the other 2-deoxy-2-fluorofuranosides.

To demonstrate the synthetic utility of this chemistry, we synthesized two fluorinated oligosaccharides, that is, 12 and 16 (Scheme 2). For the synthesis of the pentasaccharide 12 (Scheme 2a), we used the acceptor tetrasaccharide 11, which was previously prepared³² for the synthesis of an α-glucan pentasaccharide repeating unit of an immunostimulating glycan isolated from Aconitum Carmichaeli.⁴² Its reaction with the 6-O-Ac donor 3b proceeded smoothly to afford the desired pentasaccharide 12—a 2-deoxy-2-fluorogalactosyl analog of the natural pentasaccharide fragment—in 93% yield and with exclusive α-selectivity. In the synthesis of the tetrasaccharide 16 (Scheme 2b) possessing three 1,2-cis 2-deoxy-2-fluoro glycosidic linkages, the secondary acceptor methyl 2,3,6-tri-O-benzylglucoside (13) was subjected to the two-step sequence—glycosylation with the donor 3b and basic hydrolysis—to deliver the disaccharide 14 in 83% overall yield and with complete α-configuration. This sequence was repeated to afford the trisaccharide 15 with similar outcomes. In both glycosylation steps, 1.2 equiv of the donor 3b were sufficient. The final tetrasaccharide 16 was prepared from 15 in 95% yield via an S_N2 glycosylation with a 2,6-dideoxy-2-fluoroglucosyl donor. Notable in this synthetic sequence was the straightforward nature of the glycosylation steps. The entire synthesis was completed in a few days without any issues, and the absence of anomeric mixtures allowed silica gel column purification to be performed with ease.

Scheme 2 |.

Synthesis of fluorinated oligosaccharides.

To probe the directing effect of the donor phosphinoyl group, we recorded the ¹H NMR spectra of acceptor 4 at various temperatures using CD₂Cl₂ as the solvent.⁴³ At a concentration of 0.060 M, acceptor 4 displayed a clearly resolved OH triplet, with δ(OH) appearing in a narrow range of 1.65–1.90 ppm (Figure 2a). As the temperature decreased from 25 to −40 °C, the OH signal exhibited a slight but consistent downfield shift. Upon mixing acceptor 4 (0.060 M) with donor 3a (0.066 M), the OH resonance underwent a much more pronounced and increasing downfield shift as the temperature decreased (Figure 2b). A Δδ(OH) of 1.64 ppm was observed at −40 °C, indicating a strong hydrogen bond between the hydroxyl group of 4 and the phosphinoyl group of 3a under the glycosylation conditions and firmly supporting the role of such an interaction in achieving the S_N2 glycosylation.

Figure 2 |.

Partial ¹H NMR spectra in CD₂Cl₂ in variable-temperature NMR studies of acceptor 4 with or without donor 3a.

Conclusion

In conclusion, we have realized stereospecific synthesis of 2-deoxy-2-fluoroglycosides via a DGLG strategy. This S_N2 glycosylation chemistry was successfully applied to the synthesis of α−2-deoxy-2-fluoroglucose, α−2-deoxy-2-fluorogalactose, α−2,6-dideoxy-2-fluoro-glucose, α/β−2-deoxy-2-fluoroarabinose, and α/β−2-deoxy-2-fluororibose, delivering high yields and unprecedented complete stereochemical inversion at the anomeric center. These results were achieved despite the presence of a neighboring deactivating fluoro group and the potential adverse fluorine directing effect. The efficient and straightforward syntheses of complex fluorinated oligosaccharides highlight the synthetic utility of this chemistry. Despite the elaborate nature of the designed leaving group, the reaction characteristics—short reaction time (typically 1 h), resin-suitable solvent (DCM), donor access, and stereospecificity—make it versatile in sequential glycosylations and conducive to solid-phase synthesis and automated glycan assembly, in which the cost of installing such a leaving group could be tolerated. We will explore the utility of our chemistry in this direction in the future.

Supplementary Material

Supporting Information is available and includes experimental procedures and NMR spectra.