An Improved Approach for Practical Synthesis of 5-Hydroxymethyl-2′-deoxycytidine (5hmdC) Phosphoramidite and Triphosphate

5-Hydroxymethyl-2′-deoxycytidine (5hmdC) phosphoramidite and triphosphate are important building blocks in 5hmdC-containing DNA synthesis for epigenetic studies. However, efficient and practical methods for the synthesis of these compounds are still limited. The current research provides an intensively improved synthetic method that enables the preparation of commercially available cyanoethyl-protected 5hmdC phosphoramidite with an overall yield of 39% on 5 g scale. On the basis of facile and efficient accesses to cyanoethyl protected-5hmdU and 5hmdC intermediates, two efficient synthetic routes for 5hmdC triphosphate were also developed.

As shown in Figure 1A, five different types of 5hmdC phosphoramidite building blocks have been reported to date. 2-Cyanoethyl (1) is the first protecting group utilized for 5-CH 2 OH of 5hmdC [21]. Afterwards, acetyl (2) [22], cyclic carbamate (3) [23], and t-butyldimethylsilyl (4) [24] have also been employed as protecting groups. More recently, 2-nitrobenzyl-protected 5hmdC phosphoramidite (5) has enabled photocaged epigenetic labeling of DNA fragments at specific positions [25]. Currently, cyanoethyl-protected 5hmdC phosphoramidite (1) is most widely used for automated DNA synthesis and available from commercial suppliers. Though the final cleavage of cyanoethyl ether requires elongated reaction time and elevated temperature, however, its excellent stability during DNA synthesis, compatibility with existing DNA synthetic protocols, and clean deprotection are its advantages over other competitors.

Synthesis of Cyanoethyl-Protected 5hmdC Phosphoramidite (1)
As illustrated in Scheme 1, our method employed inexpensive dT starting material, which was first converted to diTBS-protected dT (7) in 96% yield and purified via simple precipitation. After free radical bromination at 5-methyl position, 5-bromomethyl dU (5BrmdU) intermediate (8) was concentrated and directly treated with 2-cyanoethanol at ambient temperature for 15 min to afford cyanoethyl-protected 5hmdU (9) in 78% yield. Treating 9 with DMT tetrafluoroborate (DMT + BF4 − ) in pyridine resulted in fast and efficient introduction of DMT to the sterically hindered 5'-OH (85%, 10). According to the new C 4 -amination method we developed, instant activation with PyAOP (1 min, 11) followed by aminolysis with conc. ammonia (4 h) yielded 5hmdC intermediate (12) in 84% yield. N-Benzoylation of 12 with benzoic anhydride in DMF at 70 °C greatly shortened reaction time of 13 from 2-4 days (at ambient temperature) [26] to only 2 h without adverse effects on either yield or chemoselectivity. Finally, phosphitylation with 2-cyanoethyl N,N-diisopropylchlorophosphoramidite in the present of triethylamine afforded the desired cyanoethyl-protected 5hmdC (1) in 83% yield. Compared to Schofield's and In 2011 and 2015, Schofield [26] and Shao [27] reported two improved methods for the synthesis of cyanoethyl-protected 5hmdC phosphoramidite (1) starting from 2'deoxyuridine (dU) and thymidine (dT), respectively ( Figure 1B). Other than the classic benzoyl (Bz) protection and phosphitylation steps, most of key steps including the initial conversion to 5-hydroxymethyl-2'-deoxyuridine (5hmdU), cyanoethyl protection, DMT protection, and C 4 -amination steps still suffer from either low-yielding or slow reaction rate. While Schofield's route (Method A) gave 1 in 24% overall yield, Shao's approach (Method B) afforded 1 in only 3% overall yield. Herein, we report an intensively optimized method for practical synthesis of 1 with an overall yield of 39% on 5 g scale. Moreover, two synthetic routes based on cyanoethyl protected-5hmdU and 5hmdC intermediates were also developed to access 5hmdC triphosphate on 50 mg scale.

Synthesis of Cyanoethyl-Protected 5hmdC Phosphoramidite (1)
As illustrated in Scheme 1, our method employed inexpensive dT starting material, which was first converted to diTBS-protected dT (7) in 96% yield and purified via simple precipitation. After free radical bromination at 5-methyl position, 5-bromomethyl dU (5BrmdU) intermediate (8) was concentrated and directly treated with 2-cyanoethanol at ambient temperature for 15 min to afford cyanoethyl-protected 5hmdU (9) in 78% yield. Treating 9 with DMT tetrafluoroborate (DMT + BF 4 − ) in pyridine resulted in fast and efficient introduction of DMT to the sterically hindered 5'-OH (85%, 10). According to the new C 4 -amination method we developed, instant activation with PyAOP (1 min, 11) followed by aminolysis with conc. ammonia (4 h) yielded 5hmdC intermediate (12) in 84% yield. N-Benzoylation of 12 with benzoic anhydride in DMF at 70 • C greatly shortened reaction time of 13 from 2-4 days (at ambient temperature) [26] to only 2 h without adverse effects on either yield or chemoselectivity. Finally, phosphitylation with 2-cyanoethyl N,Ndiisopropylchlorophosphoramidite in the present of triethylamine afforded the desired Shao's methods, the current approach achieved 39% overall yield in 5 g scale practical synthesis.
Originally, we found that 5hmdU (6) could be efficiently obtained by hydrolyzing 5BrmdU (8) with TFA/H2O (10:1, v/v) in 71% yield. However, the following synthesis of cyanoethyl ether in the presence of TFA at 100 °C according to the previous reports [26,27] yielded the desired 9 in only 51% yield. Therefore, we abandoned the synthesis of 5hmdU and subsequent condensation with cyanoethanol under harsh conditions. Instead, we attempted to convert 8 directly to cyanoethyl ether 9 via alcoholysis with cyanoethanol under mild conditions. The simultaneous alcoholysis and cleavage of TBS by HBr generated in situ only took 15 min at ambient temperature, and column purification afforded 9 in 78% yield (Scheme 2). Scheme 2. One-step synthesis of cyanoethyl-protected 5hmdU (9) from diTBS-protected 5BrhmdU (8).
In the following tritylation of 5'-OH with DMT-Cl in pyridine, we encountered the same incomplete reaction and low yielding scenario (51%, 12 h) mentioned by Schofield and coworkers [26] due to the elevated steric hindrance resulting from cyanoethyl protecting group. To solve this problem, we increased the reaction temperature to 40 and 60 °C ( Table 1, Entry 1-3). However, the yield of 10 dropped ever lower due to the formation of more diDMT-protected side product. In addition, we also tested the reactions with a catalytic amount of DMAP, 2 equiv of DIPEA, and a new tritylation reagent, DMTtrifluoroacetate (Table 1, Entry 4-7) [28]. However, none of these conditions led to improved results. Surprisingly, when DMT + BF4 − along with additives was applied according to Schofield's method, the yield of 10 was still very low ( Table 1, Entry 8) [26]. Fortunately, our further optimization found that DMT + BF4 − alone in pyridine (without any additives) could afford the desired 5'-O tritylated product 10 in 85% yield within only 1 h. A possible mechanistic explanation for this result is because pyridine favors the dissociation of Scheme 1. An improved method for practical synthesis of cyanoethyl-protected 5hmdC phosphoramidite (1).
Originally, we found that 5hmdU (6) could be efficiently obtained by hydrolyzing 5BrmdU (8) with TFA/H 2 O (10:1, v/v) in 71% yield. However, the following synthesis of cyanoethyl ether in the presence of TFA at 100 • C according to the previous reports [26,27] yielded the desired 9 in only 51% yield. Therefore, we abandoned the synthesis of 5hmdU and subsequent condensation with cyanoethanol under harsh conditions. Instead, we attempted to convert 8 directly to cyanoethyl ether 9 via alcoholysis with cyanoethanol under mild conditions. The simultaneous alcoholysis and cleavage of TBS by HBr generated in situ only took 15 min at ambient temperature, and column purification afforded 9 in 78% yield (Scheme 2). Shao's methods, the current approach achieved 39% overall yield in 5 g scale practical synthesis.
Originally, we found that 5hmdU (6) could be efficiently obtained by hydrolyzing 5BrmdU (8) with TFA/H2O (10:1, v/v) in 71% yield. However, the following synthesis of cyanoethyl ether in the presence of TFA at 100 °C according to the previous reports [26,27] yielded the desired 9 in only 51% yield. Therefore, we abandoned the synthesis of 5hmdU and subsequent condensation with cyanoethanol under harsh conditions. Instead, we attempted to convert 8 directly to cyanoethyl ether 9 via alcoholysis with cyanoethanol under mild conditions. The simultaneous alcoholysis and cleavage of TBS by HBr generated in situ only took 15 min at ambient temperature, and column purification afforded 9 in 78% yield (Scheme 2). Scheme 2. One-step synthesis of cyanoethyl-protected 5hmdU (9) from diTBS-protected 5BrhmdU (8).
In the following tritylation of 5'-OH with DMT-Cl in pyridine, we encountered the same incomplete reaction and low yielding scenario (51%, 12 h) mentioned by Schofield and coworkers [26] due to the elevated steric hindrance resulting from cyanoethyl protecting group. To solve this problem, we increased the reaction temperature to 40 and 60 °C ( Table 1, Entry 1-3). However, the yield of 10 dropped ever lower due to the formation of more diDMT-protected side product. In addition, we also tested the reactions with a catalytic amount of DMAP, 2 equiv of DIPEA, and a new tritylation reagent, DMTtrifluoroacetate (Table 1, Entry 4-7) [28]. However, none of these conditions led to improved results. Surprisingly, when DMT + BF4 − along with additives was applied according to Schofield's method, the yield of 10 was still very low ( Table 1, Entry 8) [26]. Fortunately, our further optimization found that DMT + BF4 − alone in pyridine (without any additives) could afford the desired 5'-O tritylated product 10 in 85% yield within only 1 h. A possible mechanistic explanation for this result is because pyridine favors the dissociation of Scheme 2. One-step synthesis of cyanoethyl-protected 5hmdU (9) from diTBS-protected 5BrhmdU (8).
In the following tritylation of 5'-OH with DMT-Cl in pyridine, we encountered the same incomplete reaction and low yielding scenario (51%, 12 h) mentioned by Schofield and coworkers [26] due to the elevated steric hindrance resulting from cyanoethyl protecting group. To solve this problem, we increased the reaction temperature to 40 and 60 • C ( Table 1, Entry 1-3). However, the yield of 10 dropped ever lower due to the formation of more diDMT-protected side product. In addition, we also tested the reactions with a catalytic amount of DMAP, 2 equiv of DIPEA, and a new tritylation reagent, DMT-trifluoroacetate (Table 1, Entry 4-7) [28]. However, none of these conditions led to improved results. Surprisingly, when DMT + BF 4 − along with additives was applied according to Schofield's method, the yield of 10 was still very low ( Table 1, Entry 8) [26]. Fortunately, our further optimization found that DMT + BF 4 − alone in pyridine (without any additives) could afford the desired 5'-O tritylated product 10 in 85% yield within only 1 h. A possible mechanistic explanation for this result is because pyridine favors the dissociation of DMT + BF 4 − into ionic form. Compared to the covalently linked DMT-Cl, DMT + cation is more flattened and, therefore, much easier accessible to the 5'-OH of 9. DMT + BF4 − into ionic form. Compared to the covalently linked DMT-Cl, DMT + ca more flattened and, therefore, much easier accessible to the 5'-OH of 9.  The last bottle neck of the reported synthetic routes is the conversion of 10 to counterpart 12. The reported reaction yields based on 2,4,6-triisopropylbenzenesu chloride (TPS-Cl) activating reagent showed a large discrepancy (MA: 81% vs MB In addition, both the biphasic activation and aminolysis steps were quite slow. hands, incomplete activation of urosine was observed and the yield of 11 (53%) wa to the one reported in Method B. We then tested the new BOP-based method repor Lakshman and coworkers in 2017 [29]. However, the yield of 12 was still moderate To solve this problem, we systematically investigated the reactivity of different ty onium salt coupling reagents on the C 4 amination and found that PyAOP is much advantageous in both activation and aminolysis steps than BOP, especially in the c structurally more hindered pyrimidine substrates such as 10. Compared to the T method, the ultrafast activation by PyAOP (1 min) followed by aminolysis with co trated ammonia (4 h) afforded 12 in 84% yield (Scheme 3). The more detailed con optimization, mechanistic investigation, and application scope of onium salt-bas amination method for oxidized 5mdC derivatives have been described in a previo search article reported by our group [30]. However, it should be noted that, similar above mentioned DMT + BF4 − , PyAOP is an onium salt coupling reagent. Therefo mechanistic rationale for its superb activating reactivity on sterically hindered 10 ma be ascribed to the significantly lowered steric hindrance of more flattened tripyrro phosphonium ion. The last bottle neck of the reported synthetic routes is the conversion of 10 to its dC counterpart 12. The reported reaction yields based on 2,4,6-triisopropylbenzenesulfonyl chloride (TPS-Cl) activating reagent showed a large discrepancy (MA: 81% vs MB: 58%). In addition, both the biphasic activation and aminolysis steps were quite slow. In our hands, incomplete activation of urosine was observed and the yield of 11 (53%) was close to the one reported in Method B. We then tested the new BOP-based method reported by Lakshman and coworkers in 2017 [29]. However, the yield of 12 was still moderate (56%). To solve this problem, we systematically investigated the reactivity of different types of onium salt coupling reagents on the C 4 amination and found that PyAOP is much more advantageous in both activation and aminolysis steps than BOP, especially in the case of structurally more hindered pyrimidine substrates such as 10. Compared to the TPS-Cl method, the ultrafast activation by PyAOP (1 min) followed by aminolysis with concentrated ammonia (4 h) afforded 12 in 84% yield (Scheme 3). The more detailed condition optimization, mechanistic investigation, and application scope of onium salt-based C 4 -amination method for oxidized 5mdC derivatives have been described in a previous research article reported by our group [30]. However, it should be noted that, similar to the above mentioned DMT + BF 4 − , PyAOP is an onium salt coupling reagent. Therefore, the mechanistic rationale for its superb activating reactivity on sterically hindered 10 may also be ascribed to the significantly lowered steric hindrance of more flattened tripyrrolidinophosphonium ion.
DMT + BF4 − into ionic form. Compared to the covalently linked DMT-Cl, DMT + cation is more flattened and, therefore, much easier accessible to the 5'-OH of 9.  The last bottle neck of the reported synthetic routes is the conversion of 10 to its dC counterpart 12. The reported reaction yields based on 2,4,6-triisopropylbenzenesulfonyl chloride (TPS-Cl) activating reagent showed a large discrepancy (MA: 81% vs MB: 58%). In addition, both the biphasic activation and aminolysis steps were quite slow. In our hands, incomplete activation of urosine was observed and the yield of 11 (53%) was close to the one reported in Method B. We then tested the new BOP-based method reported by Lakshman and coworkers in 2017 [29]. However, the yield of 12 was still moderate (56%). To solve this problem, we systematically investigated the reactivity of different types of onium salt coupling reagents on the C 4 amination and found that PyAOP is much more advantageous in both activation and aminolysis steps than BOP, especially in the case of structurally more hindered pyrimidine substrates such as 10. Compared to the TPS-Cl method, the ultrafast activation by PyAOP (1 min) followed by aminolysis with concentrated ammonia (4 h) afforded 12 in 84% yield (Scheme 3). The more detailed condition optimization, mechanistic investigation, and application scope of onium salt-based C 4amination method for oxidized 5mdC derivatives have been described in a previous research article reported by our group [30]. However, it should be noted that, similar to the above mentioned DMT + BF4 − , PyAOP is an onium salt coupling reagent. Therefore, the mechanistic rationale for its superb activating reactivity on sterically hindered 10 may also be ascribed to the significantly lowered steric hindrance of more flattened tripyrrolidinophosphonium ion.

Synthesis of 5hmdC Triphosphate (14)
As mentioned above, 5hmdC triphosphate (14) is the key substrate for PCR-based 5hmdC-containing DNA synthesis. Previously, we have reported a synthetic route for 14, starting from acetyl/diTBS-protected 5hmdC [31]. The current synthetic advances made in the preparation of cyanoethyl/DMT-protected 5hmdC phosphoramidite (1) offered an excellent opportunity to gain more efficient accesses to 14.
In the first synthetic route (Scheme 4), we started with cyanoethyl-protected 5hmdU (9). Activation by PyAOP (1 min) followed by aminolysis with concentrated ammonia (4 h) yielded cyanoethyl-protected 5hmdC (15) in 80% yield. However, regioselective phosphorylation at 5 -OH position gave cyanoethyl-protected 5hmdC monophosphate (16) in only 45% yield possibly due to the increased steric hindrance resulting from the cyanoethyl ether. Redox condensation with PPh 3 , 2-aminophenyl disulfide, and piperidine smoothly mediated the conversion to phosphoropiperidate 17 (6 h, 88%) [32]. The coupling of 17 and pyrophosphate was conducted with 4,5-dicyanoimidazole (DCI) via the P(V)-N activation method. Then, the cyanoethyl protecting group was removed by treating the triphosphate with concentrated ammonia at 40 • C for 12 h. As has been reported [21,26], we found that concentrated ammonia can barely cleave cyanoethyl ether at ambient temperature. However, for small molecular substrates such as 14, moderately elevated temperature (40 • C) is enough for its deprotection. The high temperature (75 • C) required in DNA synthesis is not necessary and impractical for triphosphate synthesis. Finally, purification by ion-exchange chromatography afforded the desired 5hmdC triphosphate (14) in 72% yield over two steps on 50 mg scale. Compared to our previous method based on Ac-protected 5hmdC intermediate (8 steps from diTBS-dT and 15% overall yield), the new route afforded 14 in 18% over 5 steps.

Synthesis of 5hmdC Triphosphate (14)
As mentioned above, 5hmdC triphosphate (14) is the key substrate for PCR-based 5hmdC-containing DNA synthesis. Previously, we have reported a synthetic route for 14, starting from acetyl/diTBS-protected 5hmdC [31]. The current synthetic advances made in the preparation of cyanoethyl/DMT-protected 5hmdC phosphoramidite (1) offered an excellent opportunity to gain more efficient accesses to 14.
In the first synthetic route (Scheme 4), we started with cyanoethyl-protected 5hmdU (9). Activation by PyAOP (1 min) followed by aminolysis with concentrated ammonia (4 h) yielded cyanoethyl-protected 5hmdC (15) in 80% yield. However, regioselective phosphorylation at 5′-OH position gave cyanoethyl-protected 5hmdC monophosphate (16) in only 45% yield possibly due to the increased steric hindrance resulting from the cyanoethyl ether. Redox condensation with PPh3, 2-aminophenyl disulfide, and piperidine smoothly mediated the conversion to phosphoropiperidate 17 (6 h, 88%) [32]. The coupling of 17 and pyrophosphate was conducted with 4,5-dicyanoimidazole (DCI) via the P(V)-N activation method. Then, the cyanoethyl protecting group was removed by treating the triphosphate with concentrated ammonia at 40 °C for 12 h. As has been reported [21,26], we found that concentrated ammonia can barely cleave cyanoethyl ether at ambient temperature. However, for small molecular substrates such as 14, moderately elevated temperature (40 °C) is enough for its deprotection. The high temperature (75 °C) required in DNA synthesis is not necessary and impractical for triphosphate synthesis. Finally, purification by ion-exchange chromatography afforded the desired 5hmdC triphosphate (14) in 72% yield over two steps on 50 mg scale. Compared to our previous method based on Ac-protected 5hmdC intermediate (8 steps from diTBS-dT and 15% overall yield), the new route afforded 14 in 18% over 5 steps. Due to the low-yielding monophosphorylation step in the first route, we chose to start with DMT/cyanoethyl-protected 5hmdC (12) in the second route and synthesize 14 via fully protected 5hmdC phosphoropiperidate intermediate (20) [33]. First, the 3'-OH and 4-NH2 of 14 were protected by benzyloxycarbonyl (Cbz) groups in 84% yield. After the DMT group was removed by TFA (90%, 19), 5'-OH was phosphitylated with 2-benzyl N,N-diisopropylchlorophosphoramidite. In situ hydrolysis and oxidative coupling with piperidine yielded the fully protected phosphoropiperidate 20 in 79% yield over three consecutive steps. After Cbz and Bn protecting groups were removed by catalytic hydrogenation, the deprotected phosphoropiperidate 17 was treated with pyrophosphate in the presence of DCI as the promotor. The highly efficient P(V)-N activation method afforded the triphosphate in over 95% conversion as determined by 31 P NMR. Finally, cleavage of cyanoethyl ether with concentrated ammonia at 40 °C followed by routine ion-exchange Due to the low-yielding monophosphorylation step in the first route, we chose to start with DMT/cyanoethyl-protected 5hmdC (12) in the second route and synthesize 14 via fully protected 5hmdC phosphoropiperidate intermediate (20) [33]. First, the 3'-OH and 4-NH 2 of 14 were protected by benzyloxycarbonyl (Cbz) groups in 84% yield. After the DMT group was removed by TFA (90%, 19), 5'-OH was phosphitylated with 2benzyl N,N-diisopropylchlorophosphoramidite. In situ hydrolysis and oxidative coupling with piperidine yielded the fully protected phosphoropiperidate 20 in 79% yield over three consecutive steps. After Cbz and Bn protecting groups were removed by catalytic hydrogenation, the deprotected phosphoropiperidate 17 was treated with pyrophosphate in the presence of DCI as the promotor. The highly efficient P(V)-N activation method afforded the triphosphate in over 95% conversion as determined by 31 P NMR. Finally, cleavage of cyanoethyl ether with concentrated ammonia at 40 • C followed by routine ion-exchange chromatography gave 14 in 73% yield over three consecutive steps on 50 mg scale (Scheme 5). Compared to the first synthetic route based on 9, the second route based on 20 afforded 14 in a higher overall yield (24%) but required two extra steps (total 7 steps from diTBS-dT).
Molecules 2021, 26, x FOR PEER REVIEW 6 of 11 chromatography gave 14 in 73% yield over three consecutive steps on 50 mg scale (Scheme 5). Compared to the first synthetic route based on 9, the second route based on 20 afforded 14 in a higher overall yield (24%) but required two extra steps (total 7 steps from diTBS-dT).

General Methods
Chemical reagents and solvents were purchased from Leyan-Shanghai Haohong Scientific Co. Ltd., Shanghai, China. Reactions were monitored by TLC plates coated with 0.25 mm silica gel 60 F254 and visualized by UV irradiation (254 nm). Flash column chromatography employed silica gel (particle size 32-63 μm, Qingdao Haiyang Chemicals, Qingdao, China). NMR spectra were acquired on an AV-400 instrument (Bruker BioSpin, Faellanden, Switzerland) with chemical shifts reported in ppm and referenced to CDCl3, DMSO-d6, or D2O. IR spectra were obtained with a Vertex-70 instrument (Bruker Optics, Billerica, MA, US). High-resolution mass spectra were obtained with a Dalton micrOTOF-Q II spectrometer (Bruker Optics, Billerica, MA, US) and reported as m/z. Melting points were determined with an X-4 digital melting point apparatus and uncorrected (Tech Instrument, Beijing, China). The characterization data of 1 and 9-13 and NMR spectra of all synthesized compounds are provided in the Supplementary Materials. (1) and Intermediates (9)(10)(11)(12)(13) 5-(2-Cyanoethyl)hydroxymethyl-2′-deoxyuridine (9): To a solution of 7 (9.5 g, 20.3 mmol) in dry CCl4 (100 mL) were added NBS (5.78 g, 32.5 mmol) and AIBN (541 mg, 3.3 mmol). The reaction was stirred at 75 °C for 30 min until the orange color faded. Upon cooling, the reaction solution was filtered. The filtrate was concentrated in vacuo and dried under high vacuum for 30 min. The crude 8 was dissolved in cyanoethanol (6.9 mL, 101.5 mmol) and the reaction was stirred at 20 °C for 15 min. Then NaHCO3 (3.4 g, 40.6 mmol) was added to the solution to adjust pH to 7. The solution was washed with CH2Cl2 (200 mL × 5) to remove excess cyanoethanol. Flash column chromatography of the residual syrup (CH2Cl2/MeOH = 10:1) afforded 9 (4.9 g, 78%) as a white solid.

General Methods
Chemical reagents and solvents were purchased from Leyan-Shanghai Haohong Scientific Co. Ltd., Shanghai, China. Reactions were monitored by TLC plates coated with 0.25 mm silica gel 60 F 254 and visualized by UV irradiation (254 nm). Flash column chromatography employed silica gel (particle size 32-63 µm, Qingdao Haiyang Chemicals, Qingdao, China). NMR spectra were acquired on an AV-400 instrument (Bruker BioSpin, Faellanden, Switzerland) with chemical shifts reported in ppm and referenced to CDCl 3 , DMSO-d 6 , or D 2 O. IR spectra were obtained with a Vertex-70 instrument (Bruker Optics, Billerica, MA, USA). High-resolution mass spectra were obtained with a Dalton micrOTOF-Q II spectrometer (Bruker Optics, Billerica, MA, USA) and reported as m/z. Melting points were determined with an X-4 digital melting point apparatus and uncorrected (Tech Instrument, Beijing, China). The characterization data of 1 and 9-13 and NMR spectra of all synthesized compounds are provided in the Supplementary Materials. (1) and Intermediates (9-13) 5-(2-Cyanoethyl)hydroxymethyl-2 -deoxyuridine (9): To a solution of 7 (9.5 g, 20.3 mmol) in dry CCl 4 (100 mL) were added NBS (5.78 g, 32.5 mmol) and AIBN (541 mg, 3.3 mmol). The reaction was stirred at 75 • C for 30 min until the orange color faded. Upon cooling, the reaction solution was filtered. The filtrate was concentrated in vacuo and dried under high vacuum for 30 min. The crude 8 was dissolved in cyanoethanol (6.9 mL, 101.5 mmol) and the reaction was stirred at 20 • C for 15 min. Then NaHCO 3 (3.4 g, 40.6 mmol) was added to the solution to adjust pH to 7. The solution was washed with CH 2 Cl 2 (200 mL × 5) to remove excess cyanoethanol. Flash column chromatography of the residual syrup (CH 2 Cl 2 /MeOH = 10:1) afforded 9 (4.9 g, 78%) as a white solid.   (14): Method A: To a solution of 17 (67 mg, 0.11 mmol) in DMF (2 mL) were added tris(tetra-n-butyl ammonium) hydrogen pyrophosphate (198 mg, 0.22 mmol) and DCI (78 mg, 0.66 mmol). The reaction was stirred at 20 • C for 6 h and concentrated in vacuo. The residue was dissolved in NH 3 ·H 2 O (28 w%, 5 mL), stirred at 40 • C for 12 h, and concentrated in vacuo. The residue was dissolved in aq. NaOAc solution (4 M, 0.5 mL) and then EtOH (10 mL) was added. The precipitated crude product was collected by filtration, dissolved in deionized H 2 O (0.5 mL), and loaded on a DEAE Sephadex A-25 ion exchange column (1.6 × 25 cm). Elution with NH 4 HCO 3 buffer (linear gradient 0.2 to 0.6 M), combination of appropriate fractions, and lyophilization afforded 1 in ammonium salt form. Passage of the solution of the ammonium salt in deionized H 2 O through a bed of Dowex 50W-X8 ion exchange resin (Na + form) and lyophilization afforded 1 (46 mg, 72%) as tetrasodium salt, a white solid. Method B: To a solution of 20 (106 mg, 0.13 mmol) and Et 3 N (16 µL, 0.13 mmol) in DMF (3.3 mL) was added 5 w% Pd/C (11 mg). The reaction was stirred at 20 • C under H 2 atmosphere for 3 h. The catalyst was removed by a syringe filter (0.45 µm aperture) under argon. To the DMF solution was added tris(tetra-n-butylammonium) hydrogen pyrophosphate (242 mg, 0.26 mmol) and DCI (97 mg, 0.82 mmol). The reaction was stirred at 20 • C for 6 h and concentrated in vacuo. The residue was dissolved in NH 3 ·H 2 O (28 w%, 5 mL) at 40 • C for 12 h and concentrated in vacuo. The residue was further purified according to the procedures described in Method A to afford 14 (55 mg, 73%) as tetrasodium salt, a white solid. 1

Conclusions
In summary, an improved method for practical synthesis of cyanoethtyl-protected 5hmdC phosphoramidite has been developed. Direct synthesis of cyanoethyl-protected 5hmdU via in situ alcoholysis and desilylation of 5BrmdU intermediate skipped the two limiting steps of the reported methods, i.e., the very low-yielding synthesis of 5hmdU and subsequent condensation with cyanoethanol under harsh conditions. In addition, the optimization of DMT and Bz protections along with application of the PyAOP-based C 4 amination method also significantly increased the synthetic efficiency. In a 5 g synthesis, the overall yield of 5hmdC phosphoramidite was promoted from reported 24% to 39%. In addition, two practical routes for 5hmdC triphosphate synthesis on 50 mg scale have also been developed on the basis of easily accessible cyanoethyl protected-5hmdU and 5hmdC intermediates. Moreover, mechanistic consideration of the remarkably improved reactivity of tritylating and activating reagents (DMT + BF 4 − and PyAOP) suggests that utilizing ionizable reagents could be an effective strategy to significantly promote the reactions on sterically hindered substrates due to their more flattened conformations.