3-Acetyloxy-2-cyano-2-(alkylaminocarbamoyl)propyl Groups as Biodegradable Protecting Groups of Nucleoside 5´-mono-Phosphates

Thymidine 5´-bis[3-acetyloxy-2-cyano-2-(2-phenylethylcarbamoyl)propyl]phosphate (1) has been prepared and the removal of phosphate protecting groups by hog liver carboxyesterase (HLE) at pH 7.5 and 37 °C has been followed by HPLC. The first detectable intermediates are the (RP)- and (SP)-diastereomers of the monodeacetylated triester 14, which subsequently undergo concurrent retro-aldol condensation to diester 4 and enzyme-catalyzed hydrolysis to the fully deacetylated triester 15. The former pathway predominates, representing 90% of the overall breakdown of 14. The diester 4 undergoes the enzymatic deacetylation 700 times less readily than the triester, but gives finally thymidine 5´-monophosphate as the desired main product. To elucidate the potential toxicity of the electrophilic 2-cyano-N-(2-phenylethyl)acrylamideby-product 17 released upon the deprotection, the hydrolysis of 1 has also been studied in the presence of glutathione (GSH).


Scheme 1. Structures of compounds 1-6.
The potentially toxic by-products, such as formaldehyde and electrophilic alkylating agents, released upon breakdown of the biodegradable protecting groups form a general problem of pro-drug strategies. These by-products are believed to be captured by glutathione (GSH) present in cells at a high concentration. For example, aryl vinyl ketones released from HepDirectprodrugs in hepatocytes have been shown to undergo rapid conjugation with GSH in the mice serum [34]. Glutathione has also been proposed to react with formaldehyde in human cells [35]. S-Hydroxymethylglutathione formed is oxidized by formaldehyde dehydrogenase to S-formylglutathione which is further hydrolyzed to formate by S-formylglutathione hydrolase, regenerating free GSH [36]. The secondary aim of the present study is to evaluate the alkyating ability of 2-cyano-N-(2-phenylethyl)acrylamide (17) formed upon the deprotection of 1. For this reason, the HLE-triggered deprotection in the presence of GSH has been followed by HPLC-ESI-MS. (7), prepared from ethyl cyanoacetate by acyl substitution with (2-phenylethyl)amine, was bis(hydroxymethylated) by a procedure described earlier [37]. The 2cyano-3-hydroxy-2-(hydroxymethyl)-N-(2-phenylethyl)propanamide (8) thus obtained was then converted to orthoacetate 9 and finally hydrolyzed to 3-acetyloxy-2-cyano-2-hydroxymethyl-N-(2phenylethyl)propanamide (10), essentially as described earlier (Scheme 2) [38]. Alcohol 10 was isolated and used in tetrazole promoted alcoholysis of 3´-O-levulinoylthymidine 5´-(N,Ndiethylaminophosphoramidite (12) [29] (Scheme 3). The phosphite ester obtained was oxidized with iodine in a mixture of THF, water and lutidine to the corresponding phosphate ester 13 and the levulinoyl protecting group was removed with hydrazinium acetate in a mixture of dichloromethane and MeOH.

Hydrolytic Stability of Thymidine 5´-Bis[3-acetyloxy-2-cyano-2-(2-phenylethylcarbamoyl)propyl] phosphate (1)
Hydrolysis of triester 1 was studied at pH 7.5 and 37 °C by analyzing the composition of the aliquots withdrawn from the reaction mixture at appropriate time intervals by RP HPLC. The products formed were identified by spiking with authentic samples and by mass spectrometric analysis (HPLC/ESI-MS). At pH 7.5-10, hydrolysis of the first acetic ester linkages of 1 (Reaction A in Scheme 4) was first-order in hydroxide-ion concentration. At pH 7.5, the half-life for the reaction was 28 h (k = 7.0 × 10 -6 s -1 ). The subsequent departure of the remnants of this protecting group by retro-aldol condensation gave diester 4 (Route C) without accumulation of the monodeacetylated triester 14 as an intermediate. Hydrolysis of the second ester linkage (Reaction D) was 16 times slower, the half-live being 460 h. Accordingly, the compound is slightly more stable than the corresponding 3-acetyloxymethoxy-2,2-bis(ethoxycarbonyl)propyl derived triester 2. The half-lives for the consecutive deacetylations of 2 have been reported to be 32 h and 148 h under these conditions [29]. The 3acetyloxy-2,2-bis(ethoxycarbonyl)propyl protected triester 3 and its diester counterpart 6 are considerably more stable, the half-lives for the disappearance of 3 and 6 at pH 7.5 and 25 °C being 480 h and 3850 h, respectively [29].   Table 1 records the half-lives obtained for the various partial reactions with 1 and its 3-acetyloxy-2,2-bis(ethoxycarbonyl) and 3-acetyloxymethoxy-2,2-bis(ethoxycarbonyl) counterparts 3 and 2, respectively. Although at the high HLE concentration employed the rates of Reactions B and C are comparable, Reaction C predominates at low enzyme concentrations. Comparison of the half-lives of Reaction D reveals that 1 is under such conditions converted to 5´-TMP 3 times as fast as 3, but still one order of magnitude more slowly than 2. The advantage of faster retro-aldol condensation of the intermediated derived from 1 is overcompensated by the slower enzymatic deacetylation of diester 4. In other words, the desired overall acceleration of the exposure of 5´-TMP was not achieved.
Although 2-cyano-N-(2-phenyl)ethylacrylamide (17) is largely trapped by glutathione, cyclization (Routes G and H) and hydration (Route F) of 17 seem to occur as a side reaction. As seen from Figure  4, a [M-H]ion peak at m/z 217.3, referring most likely to 2-cyano-3-hydroxy-N-(2-phenyl)ethylpropanamide (19), appears and undergoes fragmentation (loss of H 2 O) to yield an ion at m/z 199.2 (17). Similarly, the molecular ion signals of the cyclization products 20 and 21 (m/z 199.2) were detected as in the absence of GSH. By contrast, no sign of S-hydroxymethylglutathione, obtained by the reaction of GSH with formaldehyde, was observed.

General
Chemicals were purchased from Sigma-Aldrich, Fluka and Merck. Dichloromethane, acetonitrile, and pyridine were dried over 4Å molecular sieves. Dioxane was dried over 3Å molecular sieves. Triethylamine was dried by refluxing over CaH 2 and distilled before use. 1 H-, 13 C-, 31 P-NMR spectra were recorded on a Bruker Avance 400 (400 MHz for 1 H, 101 MHz for 13 C and 162 MHz for 31 P) or 500 (500 MHz for 1 H, 126 MHz for 13 C and 202 MHz for 31 P) NMR spectrometer. HRMS spectra were recorded on a Bruker Daltonics micrOTOF-Q instrument and LC-MS spectra were recorded on a Perkin-Elmer Sciex-API-365 triple-quadrupole instrument. For column chromatography, Fluka silica gel 60 (230-400 mesh) was used. Hydrolytic reactions were followed by Merck Hitachi LaChrom D7000 HPLC.

Kinetic Measurements
The reactions were carried out in sealed tubes immersed in a thermostated water bath (37.0 ± 0.1 °C). The oxonium ion concentration of the reaction solution (3.5 mL) was adjusted with glycine, 2-[4-(2-hydroxyethyl)piperazin-1-yl)]ethanesulfonic acid (HEPES) and 2-(N-morpholino)ethanesulfonic acid (MES) buffers. The ionic strength of the solutions was adjusted to 0.1 mol L -1 with sodium chloride. The hydronium ion concentration of the buffer solutions was calculated with the aid of the known pK a values of the buffer acid under the experimental conditions. The initial substrate concentration was ca. 0.4 mmol L -1 .
The enzymatic hydrolysis was carried out with Hog Liver Esterase (26 units mL -1 ) in a 2-[4-(2-hydroxyethyl)piperazin-1-yl)]ethanesulfonic acid (HEPES) buffer (0.040/0.024 mol L -1 ) at pH 7.5 and 37 °C. The samples (200 μL) withdrawn at appropriate intervals were made acidic (pH 2) with 1 mol L -1 aqueous hydrogen chloride to inactivate the enzyme and to quench the hydrolysis, cooled in an ice-bath and filtered with minisart RC 4 filters (0.45 μm). The composition of the samples was analyzed on an ODS Hypersil C18 column (4 × 250 mm 5 μm, flow rate 1 mL min -1 ), using a mixture of acetic acid/sodium acetate buffer (0.045/0.015 mol L -1 ) and MeCN, containing ammonium chloride (0.1 mol L -1 ). A good separation of the product mixtures of 2 was obtained on using a 5 min isocratic elution with the buffer containing 2% MeCN, followed by a linear gradient (23 min) up to 40.0% MeCN. Signals were recorded on a UV-detector at a wavelength of 267 nm. The reaction products were identified by the mass spectra (LC/MS) using a mixture of water and acetonitrile containing a formic acid (0.1%) as an eluent (Gemini C18 column (2 × 150 mm 5 μm, flow rate 200 μL min -1 ). The first order rate constants for the non-enzymatic hydrolysis of triester 1 were obtained by applying firstorder rate-law to the diminution of the concentration of the starting material. The enzymatic deacetylations obeyed first-order kinetics at the high HLE concentrations employed.