Synthesis and biological evaluation of fluorinated deoxynucleotide analogs based on bis-(difluoromethylene)triphosphoric acid

Abstract

It is difficult to overestimate the importance of nucleoside triphosphates in cellular chemistry: They are the building blocks for DNA and RNA and important sources of energy. Modifications of biologically important organic molecules with fluorine are of great interest to chemists and biologists because the size and electronegativity of the fluorine atom can be used to make defined structural alterations to biologically important molecules. Although the concept of nonhydrolyzable nucleotides has been around for some time, the progress in the area of modified triphosphates was limited by the lack of synthetic methods allowing to access bisCF₂-substituted nucleotide analogs—one of the most interesting classes of nonhydrolyzable nucleotides. These compounds have “correct” polarity and the smallest possible steric perturbation compared to natural nucleotides. No other known nucleotides have these advantages, making bisCF₂-substituted analogs unique. Herein, we report a concise route for the preparation of hitherto unknown highly acidic and polybasic bis(difluoromethylene)triphosphoric acid 1 using a phosphorous(III)/phosphorous(V) interconversion approach. The analog 1 compared to triphosphoric acid is enzymatically nonhydrolyzable due to substitution of two bridging oxygen atoms with CF₂ groups, maintaining minimal perturbations in steric bulkiness and overall polarity of the triphosphate polyanion. The fluorinated triphosphoric acid 1 was used for the preparation of the corresponding fluorinated deoxynucleotides (dNTPs). One of these dNTP analogs (dT) was demonstrated to fit into DNA polymerase beta (DNA pol β) binding pocket by obtaining a 2.5 Å resolution crystal structure of a ternary complex with the enzyme. Unexpected dominating effect of triphosphate/Mg²⁺ interaction over Watson–Crick hydrogen bonding was found and discussed.

Fluorine is in group VII of the periodic system, and this element should be considered, according to Pauling, a “superhalogen” (1). Fluorine is considerably more electronegative than the other halogens, and for this reason it is the only halogen that is extremely unlikely to form the positive ion. The bond energy of the C-F bond is among the highest found in natural products and is difficult to be broken enzymatically (2). Recent advances in organofluorine chemistry have been responsible for the development of a large number of new compounds of importance in biology and medicine. The knowledge gained in the synthesis of organofluorine compounds has also provided the pharmacologist with selective inhibitors of biological processes and has given the medicinal chemist the opportunity to design more active therapeutic agents. The use of fluoro compounds in studies of enzyme and pharmacological mechanisms has advantages not found with many other analogs because insight into the biochemical phenomenon can often be gained from an understanding of the altered chemical properties conferred on the compound by the fluorine substituent. The van der Waals’ radius of the fluorine atom (1.47 Å) is close to the size of hydrogen (1.2 Å). Most of the other substituent groups often used to replace hydrogen in the creation of analogs are much larger. Thus fluorine is of unique value in the design of analogs, which can very closely approach the natural biochemical intermediate. Good analogs of this kind can be useful therapeutically, but they can also be extremely valuable in defining critical sizes that contribute to structural considerations in biochemically important molecules. Further, fluorinated compounds are more hydrophobic than their hydrogen counterparts and are found to provide increase in bioavailability.

Recently, as a result of a comprehensive effort to prepare new nonhydrolyzable nucleotides, we developed a synthetic method for the preparation of the hitherto unknown bis(difluoromethylene)triphosphoric acid 1 (BMF⁴TPA, Fig. 1), a highly acidic pentabasic acid. Because of the “biostericity” and similar polarity of the CF₂ groups to the bridging oxygen atoms (3–5), this compound is a nonhydrolyzable analog of triphosphoric acid, albeit with lower pK_a^S (vide infra), a vital part of nucleotides containing both oxy- and deoxyriboses. It is important to emphasize that no other known (α,β),(β,γ)-bis-substituted triphosphate analog retains the “right” polarity and steric features of the natural triphosphate. This makes the preparation of the corresponding deoxynucleotide analogs with 1 of high scientific interest and importance. We report the synthesis and purification procedures for deoxynucleotide analogs where CF₂ groups are in both the (α,β)- and (β,γ)-positions of the triphosphate group. Owing to the unique properties of 1, these nucleotides should offer the best electronic and stereochemical mimicking of the natural deoxynucleoside triphosphates. The prepared analogs are (α,β),(β,γ)-bisCF₂ dATP, (α,β),(β,γ)-bisCF₂ dTTP, (α,β),(β,γ)-bisCF₂ dCTP, and (α,β),(β,γ)-bisCF₂ dGTP. A 2.5 Å crystallographic structure of a ternary complex of DNA pol β and (α,β),(β,γ)-bisCF₂ dTTP is also reported and discussed.

Fig. 1.

Polarities of bridging units in tiphosphate anion and its CX₂-substituted analogs.

Results and Discussion

Synthesis of BMF⁴TPA.

Although the synthesis of bis(methylene)triphosphoric acid 3 has been known for a long time (6, 7), all attempts at its direct preparation from dialkyl methylphosphonate have failed (8). Michaelis–Arbuzov-type reactions have been shown to be useful in the preparation of nonfluorinated analogs (6, 9). Although bis(difluoromethylene)triphosphoric acid 1 has a very simple structure, it remained unknown, most likely due to the failure of conventional approaches mentioned above. We have now discovered that BMF⁴TPA can be accessed starting directly from diethyl (difluoromethyl)phosphonate 4 (Scheme 1) employing phosphorous (III)/phosphorous(V) interconversion protocol.

Scheme 1.

Synthetic route to BMF⁴TPA.

Our initial approach to the synthesis of BMF⁴TPA began with the direct reaction between 2 eq. of (diethylphosphinyl)difluoromethyllithium (generated by LDA) with dichlorophosphate 5 (Scheme 1). This reaction did not provide BBMF⁴TPA ester 6; a complex mixture was formed instead. All attempts to prepare 6 with a variety of bases were unsuccessful. However, it was found that in most of the cases, the base reacted faster with the phosphorus electrophile than with phosphonate 4. This could be controlled by two factors: the rate of deprotonation and steric hindrance of the base. Generally, in the case of slow deprotonation only a small amount of (difluoromethyl)phosphonate anion was available for reaction with phosphorous electrophile, while the major part of 4 simply remained in the reaction mixture unchanged. In such a case, the majority of the base reacted with phosphorous electrophile producing the phosphoramide. These observations suggested that a sterically hindered lithium amide base would work well in the reaction, providing relatively fast deprotonation of 4 and suppression of direct reaction of the base with phosphorous electrophile. After a brief screening of several bases, it was found that use of lithium 2,2,6,6-tetramethylpiperidine amide (LTMPA) as a base gave access to compound 8 in 93% yield as determined by NMR analysis. A doublet of triplet at +6.5 ppm (2P) and a multiplet at +54.0 ppm (1P) were observed in the ³¹P NMR of the crude mixture indicating the formation of compound 8. The fluorine NMR spectrum showed two nonequivalent fluorine atoms as doublets of triplets at -116.8 and -118.5 ppm, respectively. Interestingly, the bulky lithium hexamethyldisilazide (LHMDS) did not provide positive results; a complex mixture was formed instead.

In situ oxidation of compound 8 with 2.5 equivalents of meta-chloroperbenzoic acid (m-CPBA) in dichloromethane followed by purification on silica gave access to BMF⁴TPA amido-ester 9 in 75% overall isolated yield (starting from phosphonate 4). Treatment of compound 9 with TMSBr (10) followed by hydrolysis gave access to the bis(difloromethylene)phosphoric acid 1, which was quantitatively converted to the ammonium salt by passing through DOWEX ion-exchange resin in the ammonium form. The overall yield of salt 10 was 61%.

As was discussed above, the introduction of fluorine atoms into a molecule has an impact on the physical and chemical properties of the molecule; therefore we intended to compare bond lengths and angles of the fully fluorinated analog to those of sodium triphosphate. X-ray structure of salt 10 (Fig. 2) revealed that the length of the P-O bridging bond in the original triphosphate (1.61 and 1.68 Å) is expectedly shorter than the length of the corresponding P-C bond (1.86–1.87 Å) in the fluorinated analog of the ammonium salt. Terminal P-O bond lengths for BMF⁴TPA salt were approximately the same as in sodium triphosphate (around 1.5 Å) indicating excellent isostericity for the crucial metal-binding site of the triphosphate analog. The P(2)-C(2)-P(3) angle in BMF⁴TPA was only slightly different compared to the triphosphate: 123.8° vs. 121°. However, the P(2)-C(1)-P(1) angle was significantly smaller: 114.7°; the C(1)-P(2)-C(2) angle was 106.8° vs. 98° in triphosphate. The O-P(1)-O and O-P(3)-O angles in BMF⁴TPA were approximately 114°; the O-P(1)-C(1) and O-P(3)-C(2) angles are around 102–107°, close to the value for triphosphate (106°) (11). All these data indicate that substitution of bridging oxygen atoms in triphosphate with CF₂ groups has only moderate impact on the overall spatial arrangement of triphosphate 10.

Fig. 2.

X-ray structure of BMF⁴TPA pentaamonium salt dihydrate 10.

Salt 10 can be converted back to the pentabasic acid 1 and related alkyl ammonium salts; the latter are soluble in organic solvents. We expected the lower basicity of the fluorinated analog compared to triphosphate. Indeed, the determined at 25 °C pK_a4 (5.33) and pK_a5 (7.23) of the new acid were lower than the corresponding pK_as of triphosphoric acid (5.83–6.50 and 8.73–9.24 respectively) (12, 13) with the potential to stretch the isoacidity model to new levels with the corresponding dNTP analogs.

Synthesis of dNTP Analogs.

Only a few methods for attaching low nucleophilicity phosphates to the 5′-carbon of a nucleoside are known (14, 15). These methods include the Mitsunobu reaction (8, 16–18), couplings promoted by DCC and other dehydrating agents (19, 20), electrophilic phosphorylation developed by Yoshikawa, Ludwig, and Eckstein employing 2-chloro-4H-1,3,2-benzodioxaphosphorin-4-one (21–24), nucleophilic cleavage of phosphoryl anhydride (25), reactions involving phosphoramidates (19, 26–30), Blackburn’s method employing nucleophilic substitution at 5′ position (31–33), and enzymatic phosphorylation (20, 34, 35). All of these methods lack versatility and their success depends on many factors, such as nucleophilicity/electrophilicity of the phosphate source, steric bulkiness, stability of the product under acidic or basic conditions, etc.

After comprehensive screening of several known protocols for coupling nucleosides with phosphoric, phosphonic, and carboxylic acids, we found that Balckburn’s method of nucleophilic substitution of toslylate at 5′-position with tetrabutylammonium salt 11 was the only suitable method among those screened. In the cases of deoxyadenosine, deoxythymidine, and protected deoxycytidine, conversions to 12 were 85%, 91%, and 88%, respectively. However, the protected deoxyguanosine 5′-tosylate reacted with 11 very slowly, and the maximum conversion that could be achieved was 10% (Scheme 2). Interestingly, tributylammonium salt of BMF⁴TPA 13 showed absolutely no reactivity in Blackburn’s method, although similar nucleophilic substitution with tributylammonium salt of difluoromethylene bisphosphonic acid proceeds with high yields (31–35). This could be due to intramolecular hydrogen bonding in BMF⁴TPA tributylammonium salt, which may significantly lower nucleophilicity of the anionic species.

Scheme 2.

Synthesis of deoxynucleotide analogs 12 via tosylate substitution at the 5′ position.

Biological Tests.

Having analogs 12 in hand, we demonstrated the nonhydrolyzable nature of these compounds on (α,β),(β,γ)-bisCF₂ dTTP (12b) analog as a representative example. A single-turnover gap-filling assay was performed, for correct (opposite template base dA) and incorrect (opposite template base dG) pairings (Fig. 3). After reacting for 1 min at concentrations of dNTP well above the K_d, the hydrolyzable (β,γ)-CF₂ dTTP analog (35) was incorporated to essentially 100% primer extension under both correct and mispairing conditions (lanes 4 and 8). However, no incorporation was observed after 10 min for either correct pairing or mispairing (lanes 3 and 7) when 12b was used at the same concentrations as (β,γ)-CF₂ dTTP. This result verifies that 12b is indeed nonhydrolyzable and therefore cannot be utilized by the polymerase for DNA primer extension.

Fig. 3.

Control experiment: gap-filling assay of 12b and (β,γ)-CF₂ dTTP 15.

The interesting property of dNTP analogs 12 would be their behavior in the catalytic site of a dNTP utilizing enzymes. The crystal structure of the ternary complex of DNA pol β with the incoming analog 12b opposite dA represents the precatalytic state of the nucleotidyl transfer reaction for correct incorporation, containing all atoms required for catalysis including two active site Mg²⁺ ions (PDB ID code 3LK9). As expected, the substitution of CF₂ for the (α,β)-bridging oxygen prevented dissociation of the pyrophosphate leaving group, trapping the enzyme complex prior to catalysis.

Besides the intended effect of O_α,β and O_β,γ replacement on the overall basicity of the nucleotide analog, one must also consider additional electrostatic and steric effects that may affect the active site of the enzyme. Earlier structures of DNA pol β with incoming nucleoside triphosphate analogs with CF₂ substituted for O_α,β or O_β,γ indicated that this substitution is well tolerated (35, 36). Recently, even O_β,γ has been substituted by CXY group (X,Y = H, F, Cl, Br, and /or CH₃) (37, 38). The structure of the DNA pol β precatalytic complex where both bridging oxygens are substituted with CF₂, determined to 2.5 Å resolution, superimpose well with previously determined ternary complex structures of DNA pol β Fig. 4) (35, 39). This clearly indicates that minimal structural perturbation is caused by substitution of both phosphoanhydride oxygen atoms with CF₂ groups.

Fig. 4.

Superposition of the active site of the ternary complex of DNA pol β with incoming 12b, (α,β), (β,γ)-bisCF₂ dTTP (gray), and the corresponding ternary complex with (α,β)-NH dUTP (cyan) (PDB ID code 2FMS).

Although phosphate mimics bearing difluoromethylene and fluoromethylene groups are also reported to bind in a number of other enzyme active sites (40, 41), an interesting observation was made when we measured the binding constants for the new analogs. The binding constant for the (α,β),(β,γ)-bisCF₂ dTTP analog was estimated through a steady-state inhibition analysis and found to be similar to (α,β)-CF₂-dTTP analog (K_i = 1.4 mM) (35). Likewise, the (α,β),(β,γ)-bisCF₂ dATP bound weakly (K_i = 0.7 mM). The inhibition constants were determined by competing with the corresponding natural nucleoside triphosphate (i.e., dTTP insertion opposite a templating dA in the case of bisCF₂ dTTP). When the binding constant for bisCF₂ dATP 12a was determined by competing with dTTP insertion opposite a templating dA, the binding affinity was very similar (K_i = 1.2 mM). Because Watson–Crick hydrogen bonding will not occur with the analog in this situation (i.e., would create an A–A mismatch), this result suggests that only the triphosphate portion of the analog and natural nucleotide are competing. This is also consistent with the weak binding affinity observed when the bridging O_α,β of the incoming nucleotide is substituted with CF₂ (35). Thus, although the crystal structure indicates that the bisCF₂-substituted analog is well tolerated within the polymerase active site, the primary conformation in solution does not permit Watson–Crick type hydrogen bonding.

Properties of the new set of dNTP analogs may have interesting applications in biomedical research. Although these compounds are weak inhibitors, complete nonhydrolyzability of the analogs allows very important structural evaluation of dNTP utilizing enzymes via X-ray crystallography by trapping their catalytic complexes. Low pK_a values of 1 and relatively low basicity of 12 open an opportunity for one to investigate binding pockets of other enzymes in terms of phosphate–metal interactions, which should be different especially for other than Mg²⁺ metal ions. Such studies are under way.

In conclusion, an efficient synthesis of bis(difluoromethylene)triphosphoric acid—an interesting analog of naturally abundant triphosphoric acid—was carried out. The alterations in the properties of triphosphate that can often be predicted from a fluorine substituent are valuable in the development of correlations between structure and function. As probes for the mechanism of polymerase action and its relationship to polymerase fidelity, methylene-substituted dNTP analogs permit exploration of stereoelectronic effects on active site interactions. A particularly challenging area with potentially wide practical application is the use of fluorinated analogs as probes to elucidate subtle differences in the catalytic activity of different enzymes. The triphosphate group is the part of the nucleotide molecule that undergoes catalytic cleavage; therefore its modifications may reveal additional nuances of some enzyme actions. It is hoped that our recent finding of unique fluronated deoxynucleotide analogs will give more insight into the biochemical properties of fluorinated compounds and also contribute to the understanding of the mechanism of certain types of enzyme catalysis. It seems clear that the availability of judiciously designed fluorinated nucleotide analogs is of considerable value in the development of such approaches. Moreover, a rare example of strong nonpolymeric multibasic acid 1 shall find applications in other areas of scientific research such as catalysis, analytical chemistry, fuel cell research, and medicinal chemistry.

Experimental Methods

General Remarks.

Unless otherwise mentioned, all reagents were purchased from commercial sources. Diethyl (difluoromethyl)phosphonate was prepared from chlorodifluoromethane and triethylphsophite according to a known procedure (42). Absolute THF was obtained by distillation from sodium. All NMR spectra were recorded on a 400 MHz instrument with tetramethylsilane as the ¹H standard, chlorotrifluoromethane as the ¹⁹F standard, and phosphoric acid as a standard for ³¹P experiments. The residual solvent signal (unless otherwise mentioned) was used as the standard for ¹³C NMR spectra. All reported chemical shifts are in ppm. Low- and high-resolution MS spectra and elemental microanalysis were obtained from independent commercial entities. HPLC analysis and purification of the deoxynucleotide analogs 12a and 12b were performed on a Varian ProStar 210 solvent delivery module with Shimadzu SPD-10A VP UV/Vis detector. Varian PureGel SAX 10 mm × 100 mm^-7 μm column was used for analysis (0–50% 0.5 M LiCl). Macherey-Nagel Nucleogel SAX 1000-10 25 mm × 150 mm column (0–100% 0.5 M TEAB; pH = 8.0) and Dynamax 100A C-18 21.4 mm × 250 mm column (0.1 M TEAB, 4% CH₃CN, pH >  = 7.4) were used for dual-pass preparative isolation. HPLC analysis and purification of the deoxynucleotide analogs 12c^Bz, 12c, 12d^ibu, and 12d were performed on a Shimadzu HPLC system (SCL-10A VP, SPD-10A VP, and LC-8A). Tosoh Bioscience DEAE-5PW 21.5 × 15 cm, 13 μm (100%-40% 1 M TEAB, pH = 8.0) and Shimadzu Premier C18 5 μ 250 × 23 mm preparative columns were used for purification (100%-40% 1 M TEAB, pH = 8.0). Tosoh Biosciense analytical DEAE-5PW 7.5 mm × 7.5 cm (100%-40% 1 M NaCl) and Shimadzu Premier C18 5 μ 250 × 4.6 mm columns were used for analysis.

Synthesis of [(Dimethylamino)phosphoryl] bis(difluoromethylene)phosphonate 9.

50 mL of absolute THF was placed in a 200 mL round bottom flask equipped with magnetic stirrer under Ar atmosphere. 2.84 mL (2.36 g, 16.8 mmol, 1.05 eq to the phosphonate) of 2,2,6,6-tetramethylpiperidine was added to THF, and the reaction mixture was cooled down to 0 °C. Then 6.72 mL (1.05 eq to the phosphonate) of 2.5 M n-BuLi solution in hexanes was added to the amine solution dropwise, maintaining the temperature between 0 and 5 °C. The reaction mixture was stirred at 0 °C for 30–40 min and then cooled down to -78 °C. 3 g (16 mmol) of diethyl (difluoromethyl)phosphonate 4 in 10 mL of absolute THF was added dropwise to the solution of LTMPA, maintained at -78 °C. The reaction mixture was stirred for 40–50 min at -78 °C, and then a solution of 0.915 mL (1.16 g, 7.98 mmol, 0.5 eq to the phosphonate) dimethylphosphoramidous dichloride 7 in 5 mL of absolute THF was added dropwise, maintaining the temperature at -78 °C. The reaction mixture was stirred for 2 h at -78 °C and then the temperature was slowly raised to ambient. THF was removed in vacuum, and 60 mL of dichloromethane was added. The gray heterogeneous mixture was cooled down to -78 °C, and 7.3 g (2.5 eq) of 75% m-CPBA was added portionwise. The solution was stirred for 15 min at -78 °C and was gradually warmed to 0 °C. After stirring the solution for 2 h at 0 °C, a white precipitate formed that was filtered out and washed with a small amount of dichloromethane. The organic layers were combined together, washed with saturated NaHCO₃ solution, dried over Na₂SO₄, and subjected to SiO₂ column chromatography (gradient conditions: EtOAc/Hexane = 1∶1, then EtOAc, then EtOAc/MeOH = 9.5∶0.5). The yield of compound 9 was 75%. The purity of 9 as a yellow oil may vary from 92 to 98%. ¹H NMR (CDCl₃): 1.38 (dt, J = 6.8 Hz, J = 0.8, 12H), 2.83 (s, 3H), 2.86 (s, 3H), 4.28–4.44 (m, J = 6.8 Hz, 8H) ppm. ¹³C NMR (CDCl₃): 16.5(s), 16.6(s), 35.8 (s,), 65.7 (d, J = 6.8 Hz), 65.9 (d, J = 6.8 Hz), 119.1 (m) ppm. ¹⁹F NMR (CDCl₃): -116.8 (td, J_FF = 382.0 Hz, J_FP = 82.3 Hz) ppm, -118.5 (td, J_FF = 382.0, J_FP = 82.3 Hz). ³¹P NMR (CDCl₃): 2.2–4.0 (dt, J_PF = 82.3 Hz, J_PP = 50.0 Hz, 2P), 15.8–18.4 (m, 1H) ppm. HRMS (FAB+): 466.0929 found, 466.0936 calculated for C₁₂H₂₇F₄NO₇P₃.

Synthesis of Bis(difluoromethylene)triphosphoric Acid Pentaammonium Salt 10.

2 g (4.3 mmol) of amino-ester 9 was placed in a round bottom flask equipped with a magnetic stirrer. 5.5 mL (6.5 g, 10 eq., 43 mmol) of bromotrimethylsilane was added under an Ar atmosphere, the flask was closed, covered with parafilm, and left at room temperature with stirring. After 5 d, the reaction mixture was cooled down to 0 °C and 10 mL of water was added dropwise within 2 min. The resulting biphasic solution was stirred at 0 °C for 30 min and extracted with Et₂O (3 times, 15 mL each). Then water was partially removed in vacuum, and the remaining aqueous solution was passed through DOWEX 50WX8-200 ( form) ion-exchange resin. Water was removed from the resulting solution of ammonium salt in vacuum, and the remaining solid was recrystallized from H₂O/NH₃·H₂O = 9∶1. Addition of small quantities of MeOH usually helps the crystallization process. Salt 10 was obtained as white crystals in 81% yield (after two additional collections of crystals upon evaporation of the solvent) and 61% overall yield starting from phosphonate 4. ¹⁹F NMR (D₂O): -118.3 ppm (t, J_FP = 80.5 Hz). ³¹P NMR (D₂O): 4.59 ppm (dt, J_PF = 80.5 Hz, J_PP = 40.3 Hz, 2P), 15.60–17.30 (m, 1P). MS (ESI-): 324.9 (corresponds to monodissociated BMF⁴TPA). CHN-analysis: C 5.39 (calculated 5.37). H 5.44 (calculated 5.41), N 15.68 (calculated 15.66). X-ray: see SI Appendix.

General Procedure for the Preparation of 12a–d.

0.4 mmol of 2′-deoxy-5′-tosylnucleoside 13 and 307 mg (0.2 mmol) of (Bu₄N)₅BM^F4TPA salt 11 were added to a sealable reaction vessel. 2 mL of DMF was introduced as solvent and the vessel was sealed. The mixture was heated at 110 °C for 1 hr and then left at 35 °C for several days. Conversion was checked by analytical HPLC on DEAE-5PW weak ion-exchange column with 1M NaCl as eluent. After satisfactory conversion was reached, DMF was removed under high vacuum and the remainder was dissolved in water. Undissolved particles were filtered off and the filtrate was subjected to HPLC purification on a preparative DEAE-5PW weak or SAX ion-exchange column or with 1M Et₃NHHCO₃/H₂O as eluent. Second pass through reverse phase C18 HPLC column was performed for 12a,b. Compounds 12c^Bz and 12d^ibu were deprotected in H₂O/MeOH/NH₄OH = 1∶1∶4 mixture at room temperature overnight. The next day, the reaction mixture was concentrated in vacuo, redissolved in water and subjected for reverse phase HPLC purification.

Yield of (α,β),(β,γ)-bisCF₂ dATP (12a) was 26.6 mg (24% at 100% conversion).¹H NMR (400 MHz, D₂O): 1.09 (t, J = 7.3 Hz, 27H, Et₃NH⁺), 2.59–2.69 (m, 1H), 2.83–2.93 (m, 1H), 3.07 (q, J = 7.3 Hz, 18H, Et₃NH⁺), 4.25–4.42 (m, 2H), 4.85–4.95 (m, 1H), 6.55 (t, J = 6.8 Hz, 1H), 8.31 (s, 1H), 8.61 (s, 1H) ppm. ¹⁹F NMR (376 MHz, D₂O): -116.9 (t, J = 74.1 Hz), -118.3 (t, J = 74.0 Hz) ppm. ³¹P NMR (162 MHz, D₂O): 2.5–4.5 (m, 2P), 12.0–15.0 (m, 1P) ppm. HRMS: calculated for 557.9973, found 557.9961.

Yield of (α,β),(β,γ)-bisCF₂ dTTP (12b) 19.6 mg (20% at 90% conversion).¹H NMR (400 MHz, D₂O): 1.24 (t, J = 7.3 Hz, 27H, Et₃NH⁺), 1.95 (s, 3H), 2.35–2.47 (m, 1H), 3.05 (q, J = 7.3 Hz, 18H, Et₃NH⁺ + m, 1H), 4.20–4.45 (m, 3H), 4.76 (m, 1H), 6.48 (t, J = 6.8 Hz, 1H), 7.77 (s, 1H) ppm. ¹⁹F NMR (376 MHz, D₂O): -117.7 (t, J = 74.4 Hz), -118.6 (dt, J = 74.2 Hz, J = 69.8 Hz) ppm. ³¹P NMR (162 MHz, D₂O): 2.5–4.5 (m, 2P), 12.5–15.0 (m, 1P) ppm. HRMS: calculated for 548.9858, found 548.9857.

Yield of (α,β),(β,γ)-bisCF₂ dCTP (12c) was 11.1 mg (15% at 95% conversion).¹H NMR (400 MHz, D₂O): 1.06 (t, J = 7.3 Hz, 27H, Et₃NH⁺), 2.04–2.12 (m, 1H), 2.15–2.23(m, 1H), 2.97 (q, J = 7.3 Hz, 18H, Et₃NH⁺), 3.96 (m, 1H), 4.06 (m, 2H), 4.40 (m, 1H), 5.92 (d, J = 7.6 Hz, 1H), 6.12 (t, J = 6.9 Hz, 1H), 7.79 (d, J = 7.6 Hz, 1H) ppm. ¹⁹F NMR (376 MHz, D₂O): -118.8 (t, J = 75.3 Hz), -119.3 (t, 77.9 Hz) ppm. ³¹P NMR (162 MHz, D₂O): 1.0–2.7 (m, 2P), 10.4–12.6 (m, 1P) ppm. HRMS: calculated for 533.9861, found 533.9860.

Yield of (α,β),(β,γ)-bisCF₂ dGTP (12d) was 2.2 mg (19% at 10% conversion).¹H NMR (400 MHz, D₂O): 1.09 (t, J = 7.3 Hz, 27H, Et₃NH⁺), 2.26–2.34 (m, 1H), 2.61–2.70 (m, 1H), 3.02–3.12 (m, 1H), 4.04–4.10 (m, 3H), 6.13 (dd, J = 7.8 Hz, J = 6.2 Hz, 1H), 7.91 (s, 1H) ppm. ¹⁹F NMR (376 MHz, D₂O): -118.1 (t, J = 73.0 Hz), -118.9 (t, J = 72.9 Hz) ppm. ³¹P NMR (162 MHz, D₂O): 3.6–5.4 (m, 2P), 13.0–15.0 (m, 1P) ppm. HRMS: calculated for 573.9923, found 573.9919.