Synthesis of dinucleoside acylphosphonites by phosphonodiamidite chemistry and investigation of phosphorus epimerization

Summary The reaction of the diamidite, (iPr2N)2PH, with acyl chlorides proceeds with the loss of HCl to give the corresponding acyl diamidites, RC(O)P(N(iPr)2)2 (R = Me (7), Ph (9)), without the intervention of sodium to give a phosphorus anion. The structure of 9 was confirmed by single-crystal X-ray diffraction. The coupling of the diamidites 7 and 9 with 5′-O-DMTr-thymidine was carried out with N-methylimidazolium triflate as the activator to give the monoamidites 3′-O-(P(N(iPr)2)C(O)R)-5′-O-DMTr-thymidine, and further coupling with 3′-O-(tert-butyldimethylsilyl)thymidine was carried out with activation by pyridinium trifluoroacetate/N-methylimidazole. The new dinucleoside acylphosphonites could be further oxidized, hydrolyzed to the H-phosphonates, and sulfurized to give the known mixture of diastereomeric phosphorothioates. The goal of this work was the measurement of the barrier to inversion of the acylphosphonites, which was expected to be low by analogy to the low barrier found in acylphosphines. However, the barrier was found to be high as no epimerization was detected up to 150 °C, and consistent with this, density functional theory calculations give an inversion barrier of over 40 kcal/mol.

S58 XYZ coordinates from DFT optimization of 8. NMR spectra were recorded on 400 and 500 MHz Bruker spectrometers referenced to CDCl 3 solvent peaks [1,2] and for 31 P NMR to external PPh 3 at -5.25 ppm. Peak assignments were made where possible using 2D COSY and HETCOR or HSQC spectra, with 13 C-1 H correlations shown in the spectral data where needed, as well as by comparison to the thymidine starting materials. Reaction solvents were distilled under nitrogen and then dried over activated 3 Å molecular sieves [3]. Column chromatography was carried out in the glovebox on 230-400 mesh silica gel that had been dried several hours at 250 °C under vacuum. For the peak assignments in the 1 H NMR spectra of 12 and 13, the 3-phosphorylated thymidine and the 5-phosphorylated thymidine are labeled T1 and T2, respectively.
(iPr 2 N) 2 PC(O)CH 3 (7). The starting material (iPr 2 N) 2 PH (4) was prepared via a modification of the literature procedure [4]. In the glovebox, powdered LiAlH 4 (0.242 g, 6.37 mmol, 1.17 equiv) was added in one portion to 1.451 g (5.44 mmol) of (iPr 2 N) 2 PCl [5] in 10 mL of THF, and the suspension was stirred vigorously for two hours; while it was stoppered, the stopper was removed periodically to release pressure. The grey suspension was filtered through a layer of dry Celite (CAUTION: the grey solid smokes and occasionally briefly ignites when removed from the glovebox), and the solvent was then removed from the yellow solution using a vacuum pump. The resultant white solid suspended in a yellow oil was extracted with 7  ~8 mL of hexanes, filtering each 8 mL extract through dried Celite. Solvent removal using a vacuum pump gave 1.16 g (5.00 mmol, 92% yield) of 4 as a white solid suspended in a clear oil, which on the basis of 1 H and 31 P NMR was ~93% pure. This material could be stored cold in the glovebox at -35 °C but was typically used immediately.
A sample of 4 (0.311 g, 1.34 mmol) was suspended in 7 mL CH 2 Cl 2 , the flask was fitted with a dropping funnel containing a solution of 0.119 g of CH 3

3-(5-DMTr-OT)P(N(iPr) 2 )C(O)CH 3 (10).
Solid N-methylimidazolium triflate (NMI . Tf, 0.399 g, 1.72 mmol, 0.74 equiv) [6] was added to a suspension of 7 (0.639 g, 2.33 mmol, ) and 5-O-(4,4-dimethoxytrityl)thymidine (5-DMTr-OT) [7] (1.20 g, 2.21 mmol, 0.95 equiv) in 10 mL acetonitrile. After stirring for 1.25 h, only a small amount of solid remained and the solution was filtered through Celite. Solvent removal using a vacuum pump gave a foamy yellow solid that was taken up in 10 mL benzene, 20 mL ether was added to precipitate salts, the mixture was filtered, and the solvent was again removed using a vacuum pump to give a yellow solid. This was stirred with 10 mL hexane to remove some of the starting acyl, giving 1.56 g of product as a yellow powder (99% crude yield) that was 12% starting acyl and 85% product by 31 P NMR but also contained impurities of DMTr-OT and iPr 2 NH 2 + Tf¯. Significant purification was achieved by taking up 1.32 g of this material in 8 mL benzene, filtering, and then precipitating out the product by addition of 24 mL of hexane. After cooling for 1 hr at -35 °C, the solvent was poured off and the residue pumped under vacuum to give a sticky orange solid; final solvent removal was achieved by addition of a small amount of ether and pulling a vacuum again to give a yellow foam (1.26 g, 95% recovery) that was 89% pure by 31 P NMR.
Chromatography of 0.65 g of this material on 40 mL of silica gel on a 60 mL fritted funnel, eluting with 9:1 CH 2 Cl 2 :THF, gave a yellow band collected in three fractions (60 mL 9:1 CH 2 Cl 2 :THF, 20 mL 1:1 CH 2 Cl 2 :THF and 40 mL THF); all three fractions exhibited a spot with R f = 0.5-0.55 on TLC (9:1 CH 2 Cl 2 :THF), with material at the origin eluting at the end of the last fraction. Analysis by 31 P NMR showed that the first two fractions (97.8 mg) were ~95% pure and ~86% the "fast" isomer at 117.6 ppm, while the third fraction (429 mg; 81% total recovery) was ~81% pure and ~33:67 "fast": "slow" isomers at 117.6 and 116.8 ppm. The two samples were separately rechromatographed.
The "fast" isomer was chromatographed on 10 mL of silica eluting only with 9:1 CH 2 Cl 2 :THF, and gave in the first two UV-active fractions 57.1 mg of material that was ~95% pure as a 93:7 mixture of "fast":"slow" isomers. 1
Chromatography on 30 mL of silica gel on a 60 mL fritted funnel, eluting with 9:1 CH 2 Cl 2 :THF gave a yellow band in 60 mL of solvent, discarding a pale yellow tail; solvent removal gave 0.757 g (70% crude yield) of yellow foam consisting of product and starting material. Final purification was achieved by taking up the material in 4 mL ether, and precipitating out product by addition of 10 mL of hexane with swirling, cooling briefly to -35 °C, and filtration to give the product as a yellow solid. Addition of CH 2 Cl 2 followed by solvent removal was required to remove the hexane, giving 0.632 g of yellow foam (58% yield). 1   of CH 2 Cl 2 to a 15 mL column of silica packed in 5% THF in CH 2 Cl 2 . After elution with 20 S10 mL of 5% THF in CH 2 Cl 2 followed by 10 mL of 10% THF in CH 2 Cl 2 , unidentified weakly UV-active material (6.6 mg total) eluted in 5 mL of 10% THF in CH 2 Cl 2 followed by 20 mL of 20% THF in CH 2 Cl 2 . The remaining UV-active material eluted in 30 mL of 20-30% THF in CH 2 Cl 2 followed by 20 mL of THF, giving 54.5 mg of 12 (~85% yield) that was ~84% pure by 31 P NMR and contained ~14 mol% unreacted 3-TBS-OT.
Rechromatography of this material combined with similar fractions from prior syntheses  After elution with 32 mL of 5% THF in CH 2 Cl 2 followed by 10 mL of 10% THF in CH 2 Cl 2 ,  [12] followed by TEAB gave material with only one major peak in the 31 P NMR spectrum at 113.3 ppm, and it could not be identified.
Oxidation of 13 in acetonitrile with anhydrous 3.3 M tert-butyl hydroperoxide [13] gave two peaks in the 31 P NMR spectrum for the diastereomeric oxides at -0.9 and -1.