Efficient Synthesis of UDP-Furanoses via 4,5-Dicyanoimidazole(DCI)-Promoted Coupling of Furanosyl-1-Phosphates with Uridine Phosphoropiperidate

A P(V)-N activation method based on nucleoside phosphoropiperidate/DCI system has been developed for improved synthesis of diverse UDP-furanoses. The reaction conditions including temperature, amount of activator, and reaction time were optimized to alleviate the degradation of UDP-furanoses to cyclic phosphates. In addition, an efficient and facile phosphoramidite route was employed for the preparation of furanosyl-1-phosphates.

of furanosyl-1-phosphate with activated UMPs, such as imidazolide [22][23][24], morpholidate [25,26], N-methyl imidazolide [27,28], has been utilized as the major approach to afford UDP-furanoses. The reactions of imidazolide and morpholidate are typically low-yielding (<20%) and require 1-2 days to finish. Although the employment of more reactive N-methyl imidazolide shortens the reaction time to a few hours, the yields of UDP-furanoses are low (8-35%). Alternatively, benzimidazolyl thiofuranosides have been prepared to couple with UDP [29]. The reaction rate of this method is fast (10 min-1 h). But the yields of UDP-furanoses are still within the range of 20-30%. In addition, the benzimidazolyl thiofuranosides have also been employed for the preparation of furanosyl-1phosphates. The combined strength of chemical synthesis of furanosyl-1-phosphates and a galactose-1phosphate uridylyltransferase provided a stereoselective chemoenzymatic approach to specific UDP-furanoses [30].
In our previous research, we developed a novel P(V)-N activation strategy based on nucleoside 5 -phosphoropiperidate/4,5-dicyanoimidazole (DCI) system for highly efficient synthesis of nucleoside 5 -polyphosphates [31][32][33]. Its further application on the synthesis of NDP-6-deoxy-L-pyranoses showed that piperidate/DCI system is much more reactive than the conventional imidazolide or morpholidate/1H-tetrazole system [34]. Herein, we report an improved synthesis of diverse UDP-furanoses based on the piperidate/DCI system. An efficient and facile phosphoramidite route for the preparation of furanosyl-1-phosphates was also developed.

A Phosphoramidite Approach for Furanosyl-1-Phosphate Synthesis
For the P(V)-N activation strategy-based synthesis of UDP-furanoses, sugar-1-phosphate is an essential reactant but with poor availability. Previously, furanosyl-1-phosphates have been primarily prepared by coupling benzoyl-protected furanosyl-1-bromide with dibenzyl phosphate [20]. This method typically requires long reaction time and the yields are moderate (40-50%). Alternatively, furanosyl-1-phosphates have been prepared from thiofuranosides [30,35]. Based on our previous research on the phosphorylation of saccharides and nucleosides [36,37], we attempted to synthesize furanosyl-1-phosphates via a phosphoramidite route. As shown in Scheme 1, phosphitylation of 1-OH of Bz-protected furanoses 1-4 with benzyl-N,N-diisopropylchlorophosphoramidite, 1H-tetrazolecatalyzed alcoholysis, and in situ oxidation with H 2 O 2 afforded crude 5-8 in good yields over three consecutive fast steps. However, we found that the α anomer of L-Araf (5) decomposed completely during flash column chromatography and only the β anomer was isolated in 32% yield. Similarly, only the β anomer of D-Galf (6) survived column purification and was obtained in 30% yield. In contrast, the phosphate triesters of D-Glcf (7) and D-6F-Galf (8) were much more stable and isolated as anomeric mixtures in over 75% yields. Catalytic hydrogenation followed by treatment with MeOH/H 2 O/Et 3 N (5:2:1) gave the desired furanosyl-1-phosphates 9-12 in excellent yields. UDP-furanoses. The reactions of imidazolide and morpholidate are typically low-yielding (<20%) and require 1-2 days to finish. Although the employment of more reactive N-methyl imidazolide shortens the reaction time to a few hours, the yields of UDP-furanoses are low (8-35%). Alternatively, benzimidazolyl thiofuranosides have been prepared to couple with UDP [29]. The reaction rate of this method is fast (10 min-1 h). But the yields of UDP-furanoses are still within the range of 20-30%. In addition, the benzimidazolyl thiofuranosides have also been employed for the preparation of furanosyl-1-phosphates. The combined strength of chemical synthesis of furanosyl-1-phosphates and a galactose-1-phosphate uridylyltransferase provided a stereoselective chemoenzymatic approach to specific UDP-furanoses [30]. In our previous research, we developed a novel P(V)-N activation strategy based on nucleoside 5′-phosphoropiperidate/4,5-dicyanoimidazole (DCI) system for highly efficient synthesis of nucleoside 5′-polyphosphates [31][32][33]. Its further application on the synthesis of NDP-6-deoxy-Lpyranoses showed that piperidate/DCI system is much more reactive than the conventional imidazolide or morpholidate/1H-tetrazole system [34]. Herein, we report an improved synthesis of diverse UDP-furanoses based on the piperidate/DCI system. An efficient and facile phosphoramidite route for the preparation of furanosyl-1-phosphates was also developed.

A Phosphoramidite Approach for Furanosyl-1-Phosphate Synthesis
For the P(V)-N activation strategy-based synthesis of UDP-furanoses, sugar-1-phosphate is an essential reactant but with poor availability. Previously, furanosyl-1-phosphates have been primarily prepared by coupling benzoyl-protected furanosyl-1-bromide with dibenzyl phosphate [20]. This method typically requires long reaction time and the yields are moderate (40-50%). Alternatively, furanosyl-1-phosphates have been prepared from thiofuranosides [30,35]. Based on our previous research on the phosphorylation of saccharides and nucleosides [36,37], we attempted to synthesize furanosyl-1-phosphates via a phosphoramidite route. As shown in Scheme 1, phosphitylation of 1-OH of Bz-protected furanoses 1-4 with benzyl-N,N-diisopropylchlorophosphoramidite, 1Htetrazole-catalyzed alcoholysis, and in situ oxidation with H2O2 afforded crude 5-8 in good yields over three consecutive fast steps. However, we found that the α anomer of L-Araf (5) decomposed completely during flash column chromatography and only the β anomer was isolated in 32% yield. Similarly, only the β anomer of D-Galf (6) survived column purification and was obtained in 30% yield. In contrast, the phosphate triesters of D-Glcf (7) and D-6F-Galf (8) were much more stable and isolated as anomeric mixtures in over 75% yields. Catalytic hydrogenation followed by treatment with MeOH/H2O/Et3N (5:2:1) gave the desired furanosyl-1-phosphates 9-12 in excellent yields. Scheme 1. A phosphoramidite method for the synthesis of furanosyl-1-phosphates 9-12. Scheme 1. A phosphoramidite method for the synthesis of furanosyl-1-phosphates 9-12. 2.2. The Attempt to Synthesize UDP-L-Araf (16) from Benzoyl-Protected L-Arabinofuranosyl-1-Phosphate (13) As mentioned above, UDP-furanoses are easy to decompose to 1,2-cyclic furanosyl phosphate due to the presence of free 2-OH. To avoid the formation of 1,2-cyclic phosphate byproduct, our initial attempt to synthesize UDP-furanoses via the P(V)-N activation strategy was to couple 2,3,5-tribenzoylprotected L-Araf -1-phosphate (13) with 2 equiv of uridine 5 -phosphoropiperidate (14) in the presence of 2 equiv of DCI in DMF at 30 • C. 31 P NMR tracing experiment (Figure 1) showed that Bz-protected UDP-L-Araf (15) formed smoothly over 24 h when piperidate 14 was completely consumed. Whilẽ 10% of the starting sugar phosphate 13 was left unreacted, its conversion to 15 was nearly 90% based on 31 P NMR integration. Meanwhile, diuridine diphosphate (Up 2 U) and uridine monophosphate (UMP) were observed as the major byproducts. It is noteworthy that benzoyl-protection on L-Araf moiety completely suppressed the formation of 1,2-cyclic phosphate byproduct and the desired 15 was isolated in 71% yield.  (16) from Benzoyl-Protected L-Arabinofuranosyl-1-Phosphate (13) As mentioned above, UDP-furanoses are easy to decompose to 1,2-cyclic furanosyl phosphate due to the presence of free 2-OH. To avoid the formation of 1,2-cyclic phosphate byproduct, our initial attempt to synthesize UDP-furanoses via the P(V)-N activation strategy was to couple 2,3,5tribenzoyl-protected L-Araf-1-phosphate (13) with 2 equiv of uridine 5′-phosphoropiperidate (14) in the presence of 2 equiv of DCI in DMF at 30 °C. 31 P NMR tracing experiment (Figure 1) showed that Bz-protected UDP-L-Araf (15) formed smoothly over 24 h when piperidate 14 was completely consumed. While ~10% of the starting sugar phosphate 13 was left unreacted, its conversion to 15 was nearly 90% based on 31 P NMR integration. Meanwhile, diuridine diphosphate (Up2U) and uridine monophosphate (UMP) were observed as the major byproducts. It is noteworthy that benzoyl-protection on L-Araf moiety completely suppressed the formation of 1,2-cyclic phosphate byproduct and the desired 15 was isolated in 71% yield. Although the P(V)-N activation strategy yielded an excellent result in the coupling step, we were also clearly aware of the possibility that the final deprotection step may cause undesired decomposition to sugar 1,2-cyclic phosphate. As shown in Scheme 2, both MeOH/H2O/Et3N (7:3:1) [38] and MeONa/MeOH [36] deprotection methods caused almost total breakdown of 15 to equal amounts of 1,2-cyclic phosphate and UMP, while the MeOH/TEAB buffer/Et3N (3:4:0.05) method reported by Williams et al. [39] was not effective at the claimed low temperature (−20 °C). The above experimental results reveal that the UDP-L-Araf has very strong tendency to decompose via the intramolecular cyclization under basic conditions at room temperature.  Although the P(V)-N activation strategy yielded an excellent result in the coupling step, we were also clearly aware of the possibility that the final deprotection step may cause undesired decomposition to sugar 1,2-cyclic phosphate. As shown in Scheme 2, both MeOH/H 2 O/Et 3 N (7:3:1) [38] and MeONa/MeOH [36] deprotection methods caused almost total breakdown of 15 to equal amounts of 1,2-cyclic phosphate and UMP, while the MeOH/TEAB buffer/Et 3 N (3:4:0.05) method reported by Williams et al. [39] was not effective at the claimed low temperature (−20 • C). The above experimental results reveal that the UDP-L-Araf has very strong tendency to decompose via the intramolecular cyclization under basic conditions at room temperature. 2.2. The Attempt to Synthesize UDP-L-Araf (16) from Benzoyl-Protected L-Arabinofuranosyl-1-Phosphate (13) As mentioned above, UDP-furanoses are easy to decompose to 1,2-cyclic furanosyl phosphate due to the presence of free 2-OH. To avoid the formation of 1,2-cyclic phosphate byproduct, our initial attempt to synthesize UDP-furanoses via the P(V)-N activation strategy was to couple 2,3,5tribenzoyl-protected L-Araf-1-phosphate (13) with 2 equiv of uridine 5′-phosphoropiperidate (14) in the presence of 2 equiv of DCI in DMF at 30 °C. 31 P NMR tracing experiment ( Figure 1) showed that Bz-protected UDP-L-Araf (15) formed smoothly over 24 h when piperidate 14 was completely consumed. While ~10% of the starting sugar phosphate 13 was left unreacted, its conversion to 15 was nearly 90% based on 31 P NMR integration. Meanwhile, diuridine diphosphate (Up2U) and uridine monophosphate (UMP) were observed as the major byproducts. It is noteworthy that benzoyl-protection on L-Araf moiety completely suppressed the formation of 1,2-cyclic phosphate byproduct and the desired 15 was isolated in 71% yield. Although the P(V)-N activation strategy yielded an excellent result in the coupling step, we were also clearly aware of the possibility that the final deprotection step may cause undesired decomposition to sugar 1,2-cyclic phosphate. As shown in Scheme 2, both MeOH/H2O/Et3N (7:3:1) [38] and MeONa/MeOH [36] deprotection methods caused almost total breakdown of 15 to equal amounts of 1,2-cyclic phosphate and UMP, while the MeOH/TEAB buffer/Et3N (3:4:0.05) method reported by Williams et al. [39] was not effective at the claimed low temperature (−20 °C). The above experimental results reveal that the UDP-L-Araf has very strong tendency to decompose via the intramolecular cyclization under basic conditions at room temperature. product. As expected, the formation of 1,2-cyclic phosphate was observed. Surprisingly, the reaction between 14 and fully deprotected 9 was significantly faster (8 h) than that with Bz-protected 13, which in turn reduced the undesired self-condensation of 14. On the basis of 31 P NMR integration, it was estimated that UDP-L-Araf (16) and 1,2-cyclic phosphate were obtained in 61% and 25% yields, respectively.

The P(V)-N Activation Method for the Synthesis UDP-Furanoses from Furanosyl-1-Phosphates
The failure to deprotect 15 without causing decomposition urged us to go back to the direct coupling of phosphate 9 with piperidate 14 under the P(V)-N activation conditions ( Figure 2). The reaction temperature was intentionally lowered to 20 °C to alleviate degradation of the UDP-L-Araf product. As expected, the formation of 1,2-cyclic phosphate was observed. Surprisingly, the reaction between 14 and fully deprotected 9 was significantly faster (8 h) than that with Bz-protected 13, which in turn reduced the undesired self-condensation of 14. On the basis of 31 P NMR integration, it was estimated that UDP-L-Araf (16) and 1,2-cyclic phosphate were obtained in 61% and 25% yields, respectively. To suppress the degradation of 16 and self-condensation of 14, we further reduced the amount of activating reagent. When 1.25 equiv of DCI was applied, close analysis of the composition of the reaction mixture at different reaction time showed that the yield of 16 reached the maximum level (68%) at 8 h when 22% of 9 remained unreacted and 10% 1,2-cyclic phosphate byproduct was formed ( Figure 3). However, as the reaction proceeded, the yield of 16 did not increase anymore due to its degradation, which occurred in a comparable rate to its formation. When most of 9 was consumed (5% left) at 18 h, the formation of 1,2-cyclic phosphate (increased to 27%) and Up2U byproduct was further aggravated. Therefore, quenching the reaction at 8 h should be beneficial for easier purification. With the optimized conditions, four UDP-furanoses 16-19 were prepared via the P(V)-N activation method. The coupling of furanosyl-1-phosphates 9-12 with 2 equiv of piperidate 14 in the To suppress the degradation of 16 and self-condensation of 14, we further reduced the amount of activating reagent. When 1.25 equiv of DCI was applied, close analysis of the composition of the reaction mixture at different reaction time showed that the yield of 16 reached the maximum level (68%) at 8 h when 22% of 9 remained unreacted and 10% 1,2-cyclic phosphate byproduct was formed ( Figure 3). However, as the reaction proceeded, the yield of 16 did not increase anymore due to its degradation, which occurred in a comparable rate to its formation. When most of 9 was consumed (5% left) at 18 h, the formation of 1,2-cyclic phosphate (increased to 27%) and Up 2 U byproduct was further aggravated. Therefore, quenching the reaction at 8 h should be beneficial for easier purification.

The P(V)-N Activation Method for the Synthesis UDP-Furanoses from Furanosyl-1-Phosphates
The failure to deprotect 15 without causing decomposition urged us to go back to the direct coupling of phosphate 9 with piperidate 14 under the P(V)-N activation conditions (Figure 2). The reaction temperature was intentionally lowered to 20 °C to alleviate degradation of the UDP-L-Araf product. As expected, the formation of 1,2-cyclic phosphate was observed. Surprisingly, the reaction between 14 and fully deprotected 9 was significantly faster (8 h) than that with Bz-protected 13, which in turn reduced the undesired self-condensation of 14. On the basis of 31 P NMR integration, it was estimated that UDP-L-Araf (16) and 1,2-cyclic phosphate were obtained in 61% and 25% yields, respectively. To suppress the degradation of 16 and self-condensation of 14, we further reduced the amount of activating reagent. When 1.25 equiv of DCI was applied, close analysis of the composition of the reaction mixture at different reaction time showed that the yield of 16 reached the maximum level (68%) at 8 h when 22% of 9 remained unreacted and 10% 1,2-cyclic phosphate byproduct was formed ( Figure 3). However, as the reaction proceeded, the yield of 16 did not increase anymore due to its degradation, which occurred in a comparable rate to its formation. When most of 9 was consumed (5% left) at 18 h, the formation of 1,2-cyclic phosphate (increased to 27%) and Up2U byproduct was further aggravated. Therefore, quenching the reaction at 8 h should be beneficial for easier purification. With the optimized conditions, four UDP-furanoses 16-19 were prepared via the P(V)-N activation method. The coupling of furanosyl-1-phosphates 9-12 with 2 equiv of piperidate 14 in the With the optimized conditions, four UDP-furanoses 16-19 were prepared via the P(V)-N activation method. The coupling of furanosyl-1-phosphates 9-12 with 2 equiv of piperidate 14 in the presence of 1.25 equiv of DCI at 20 • C for 8 h afforded 16-19 in 49-53% yields after preparative HPLC purification (Scheme 3). It was determined that adjustment of the pH of TEAB buffer (10 mM) to 8.0 by adding acetic acid is helpful to minimize the decomposition of UDP-furanoses during HPLC purification. It is worth mentioning that 31 P NMR tracing of the reactions of 18 and 19 showed that the anomeric configuration of furanosyl is an important factor for not only the stability of the UDP-furanose product but also the reactivity of the sugar-1-phosphate reactant. For instance, the UDP-α-D-Glcf was much labile than UDP-β-D-Glcf and completely decomposed during the reaction. When the reaction of D-6F-Galf -1-phosphate (12) was stopped at 8 h, 95% of the β anomer had been converted to the product, but only 68% of the α anomer had reacted. After preparative HPLC purification, only UDP-β-D-6F-Galf (19) was obtained as the sole product.
presence of 1.25 equiv of DCI at 20 °C for 8 h afforded 16-19 in 49-53% yields after preparative HPLC purification (Scheme 3). It was determined that adjustment of the pH of TEAB buffer (10 mM) to 8.0 by adding acetic acid is helpful to minimize the decomposition of UDP-furanoses during HPLC purification. It is worth mentioning that 31 P NMR tracing of the reactions of 18 and 19 showed that the anomeric configuration of furanosyl is an important factor for not only the stability of the UDPfuranose product but also the reactivity of the sugar-1-phosphate reactant. For instance, the UDP-α-D-Glcf was much labile than UDP-β-D-Glcf and completely decomposed during the reaction. When the reaction of D-6F-Galf-1-phosphate (12) was stopped at 8 h, 95% of the β anomer had been converted to the product, but only 68% of the α anomer had reacted. After preparative HPLC purification, only UDP-β-D-6F-Galf (19) was obtained as the sole product.

General Methods
General chemical reagents and solvents were obtained from commercial suppliers. The Bzprotected furanoses (1-4) were prepared according to known procedures [27,29,[40][41][42]. Uridine 5′phosphoropiperidate (14) was prepared according to a previous report [32]. All reactions were performed under an atmosphere of inert gas and monitored by thin layer chromatography on plates coated with 0.25 mm silica gel 60 F254. TLC plates were visualized by UV irradiation (254 nm). Flash column chromatography employed silica gel (particle size 32-63 μm). All NMR spectra were obtained with a Bruker AV-400 instrument (Billerica, MA, USA) with chemical shifts reported in parts per million (ppm, δ) and referenced to CDCl3, MeOH-d4, or D2O. The NMR spectra of compounds 5-13 and 15-19 were provided in Supplementary Materials (Figures S1-S42). Low-and high-resolution mass spectra were reported as m/z and obtained with a Bruker amaZon SL and a Bruker Dalton microTOFQ II mass spectrometer, respectively. The HPLC traces of 16-19 were recorded on an Agilent 1260 instrument equipped with a Waters XTerra MS C18 analytical column (4.6 × 150 mm, 5 μm) [flow rate = 1.0 mL/min; linear gradient of 5% to 100% MeOH in TEAB buffer (10 mM, pH 8.0) over 10 min; UV detection at 254 nm] and provided in Supplementary Materials (Figures S43-S46).

General Methods
General chemical reagents and solvents were obtained from commercial suppliers. The Bz-protected furanoses (1-4) were prepared according to known procedures [27,29,[40][41][42]. Uridine 5 -phosphoropiperidate (14) was prepared according to a previous report [32]. All reactions were performed under an atmosphere of inert gas and monitored by thin layer chromatography on plates coated with 0.25 mm silica gel 60 F254. TLC plates were visualized by UV irradiation (254 nm). Flash column chromatography employed silica gel (particle size 32-63 µm). All NMR spectra were obtained with a Bruker AV-400 instrument (Billerica, MA, USA) with chemical shifts reported in parts per million (ppm, δ) and referenced to CDCl 3 , MeOH-d 4 , or D 2 O. The NMR spectra of compounds 5-13 and 15-19 were provided in Supplementary Materials (Figures S1-S42). Low-and high-resolution mass spectra were reported as m/z and obtained with a Bruker amaZon SL and a Bruker Dalton microTOFQ II mass spectrometer, respectively. The HPLC traces of 16-19 were recorded on an Agilent 1260 instrument equipped with a Waters XTerra MS C18 analytical column (4.6 × 150 mm, 5 µm) [flow rate = 1.0 mL/min; linear gradient of 5% to 100% MeOH in TEAB buffer (10 mM, pH 8.0) over 10 min; UV detection at 254 nm] and provided in Supplementary Materials (Figures S43-S46).

General Synthetic Procedure and Characterization of Furanosyl-1-Phosphates 9-13
To a solution of phosphite triesters 5-8 (200 mg) and 2 equiv of Et 3 N in dry methanol (3 mL) was added 5 wt% Pd/C (20 mg). The reaction was stirred under an atmosphere of hydrogen at ambient temperature for 8 h. Then, the catalyst was filtered off. The filtrate was dried to afford crude phosphate. For compound 13, the crude product was directly dissolved in CH 2 Cl 2 and purified by Sephadex LH-20 size exclusion gel chromatography. For compounds 9-12, the crude products were dissolved in MeOH/H 2 O/Et 3 N (5:2:1, 10 mL) and stirred for 24 h at ambient temperature. After concentration, the crude product was dissolved in MeOH and purified by Sephadex LH-20 size exclusion gel chromatography. A combination of appropriate fractions and concentration afforded 9-13 in triethylammonium salt form.

Conclusions
In summary, our attempt to synthesize UDP-furanoses via the phosphoropiperidate/DCI system-based P(V)-N activation strategy afforded a more efficient approach. Due to the labile nature of UDP-furanoses, reaction conditions such as temperature (20 • C), amount of activator (1.25 equiv), and reaction time (8 h) were optimized to alleviate the degradation of UDP-furanoses to sugar 1,2-cyclic phosphates. In addition, a phosphoramidite approach for the preparation of furanosyl-1-phosphates, which involves three consecutive fast steps, was also developed. Although the deprotection of benzoyl groups of UDP-L-Araf precursor 15 caused severe degradation to L-Araf -1,2cyclic phosphate, the much higher coupling efficacy of protected sugar-1-phosphate and complete absence of product degradation exhibited a huge advantage over the coupling with fully deprotected furanosyl-1-phosphates. Future discovery of an efficient and selective deprotection method for benzoyl groups could greatly improve the chemical synthesis of UDP-furanoses.