Synthesis of Fluorine-Containing Analogues of Purine Deoxynucleosides: Optimization of Enzymatic Transglycosylation Conditions

In this work, a comparative analysis of the conditions of transglycosylation reactions catalyzed by E. coli nucleoside phosphorylases was carried out, and the optimal conditions for the formation of various nucleosides were determined. Under the optimized conditions of transglycosylation reaction, fluorine-containing derivatives of N6-benzyl-2'-deoxyadenosine, potential inhibitors of replication of enteroviruses in a cell, were obtained starting from the corresponding ribonucleosides.


INTRODUCTION
Enzymatic transglycosylation methods are widely used for obtaining drugs on the basis of nucleosides and their analogues and are based on the reaction of transferring a carbohydrate residue from one heterocyclic base to another [1][2][3]. Nucleoside phosphorylases (NP), which perform reversible phosphorolysis of ribonucleosides/2'-deoxyribonucleosides with the formation of the corresponding heterocyclic base and α-D-(2-deoxy)ribofuranose-1-phosphate ((d)Rib-P). The equilibrium of the phosphorolysis reaction is shifted towards the formation of nucleosides, and in the case of purines it is more significant [1][2][3][4][5][6], which makes it possible to use two coupled reactions of phosphorolysis, a donor nucleoside and a nucleoside containing a heterocyclic base-acceptor, to perform an enzymatic transglycosylation reaction, during which a carbohydrate residue is transferred from a pyrimidine or purine nucleoside donor to a purine heterocyclic base acceptor (Fig. 1). This general scheme makes it possible to obtain new modified nucleosides depend-ing on the set of used starting compounds and the substrate specificity of NP.
Earlier, approaches to optimizing the transglycosylation reaction using 7-methyl-2'-deoxyguanosine as the starting substrate for the production of α-D-2deoxyribose-1-phosphate (dRib-1-P), 5-substituted derivatives of 2'-deoxyuridine, cladribine, and allopurinol-riboside [5,7], were studied in the Laboratory of Design and Synthesis of Biologically Active Compounds (DSBAC) of the Engelhard Institute of Molecular Biology, Russian Academy of Sciences. A mathematical model of the transglycosylation process was proposed, which can be used to quantify the effect of initial conditions on the result of transglycosylation [8]. This work is a continuation of earlier studies aimed at expanding knowledge about the substrate specificity of NP and obtaining modified analogues of natural nucleosides by enzymatic transglycosylation.

DISCUSSION
In this work, a comparative analysis of the transglycosylation reaction involving purine (PNP) and pyrimidine (UP, TP) E. coli nucleoside phosphorylases was performed ( Table 1). The selection of enzymes of bacterial origin as catalysts was determined by their wide substrate specificity, pH optimum in neutral/weakly alkaline media, and a fairly wide operating temperature range, which allows the reaction to be carried out under mild conditions without notice- able nonspecific cleavage of the N-glycosidic bond with yields of target products close to the theoretically predicted ones [8][9][10][11][12][13].
According to Fig. 1, the transglycosylation reaction proceeds through the formation of α-D-ribose-1phosphate (Rib-1-P) or α-D-2-deoxyribose-1-phosphate (dRib-1-P). The synthesis of purine nucleosides from pyrimidine nucleosides and vice versa requires the participation of two enzymes: PNP and UP (or TP). The equilibrium constants of phosphorolysis of natural pyrimidine nucleosides in the presence of UP and TP are higher than the equilibrium constants of phosphorolysis of natural purine nucleosides in the presence of PNP [2,5,8,11,14]. Therefore, it is more reasonable to use Rib-1-P and dRib-1-P (or pyrimidine, but not purine, nucleosides) as donors [8,15,16].
This expediency is well confirmed by the experimental data presented in Table 1.
When adenosine was obtained from uridine, the reaction proceeded with a high yield, which was determined as the ratio of the equilibrium concentration of the product to the initial concentration of the starting base or glycosyl donor, depending on what was taken in deficiency (Table 1, lines 1-3, 77-94% according to HPLC). When uridine was obtained from adenosine, the reaction yield was significantly reduced ( Table 1, lines 4-6, 22-27% according to HPLC). In the series of purine nucleosides, inosine was a more productive glycosyl donor than adenosine; therefore, the reaction for obtaining adenosine from inosine proceeded with a higher yield ( . With an increase in the amount of the donor nucleoside in the reaction mixture, it is naturally possible to increase the yield of the target nucleoside (and, accordingly, the conversion of the base); however, this also increases the number of unreacted components, which increases the complexity of the subsequent processing of the reaction mixture and further purification (Table 1, lines 3, 6, 13, 15).
The transglycosylation reaction can be simplified in two ways: (1) exclusion of the stage of phosphorolysis of the nucleoside that serves as a glycosyl donor in the transglycosylation reaction ( Fig. 1, stage 1) by introducing a ready-to-use Rib-1-P or dRib-1-P into the reaction [7]; (2) transformation of stage 1 into an irreversible one using 7-methyl-(2'-deoxy)guanosine (7-Me(d)Guo) as a source of ribose residue due to its almost irreversible phosphorolysis [17,18].
The use of (d)Rib-1-P reduces the number of components in the reaction mixture, facilitates the isolation of target compounds, and makes it possible to significantly shift the equilibrium of the glycosylation reaction towards the formation of nucleosides (Table 1, line 14, 98% according to HPLC; the yield was calculated from the reagent taken in deficiency (Rib-1-P)). The replacement of (d)Rib-1-P with the more accessible 7-Me(d)Guo leads to comparable yields of the reaction products (Table 1).
The synthesis method consisted of three separate stages, each stage was catalyzed by E. coli PNP. The starting bases N 6 -(3-trifluoromethylbenzyl)adenine (TFMBn-Ade, 2a) and N 6 -pentafluorophenylmethyladenine (PFPh-Ade, 2b) were obtained from ribonucleosides under the conditions of enzymatic arsenolysis (Fig. 2, stage 1). Enzymatic arsenolysis is based on the cleavage of ribonucleoside in the presence of potassium dihydroorthoarsenate (KH 2 AsO 4 ) into a purine base and the highly labile α-D-ribofuranose-1-arsenate (Rib-1-As), which is irreversibly hydrolyzed; this shifts the equilibrium of ribonucleoside cleavage towards the formation of a base [20]. The poor solubility of heterocyclic bases 2a and 2b in water and in Tris-HCl buffer also leads to a shift in equilibrium towards the formation of products. To prevent the formation of a mixture of ribo-and deoxyribonucleosides during further stages, stage 1 should be per-formed in a separate flask. Then, bases 2a and 2b were filtered off and introduced into the transglycosylation reaction with 7-Me-dGuo in the presence of potassium dihydroorthophosphate and E. coli PNP (Fig. 2, stages 2-3). In the reaction mixture, 7-Me-dGuo was converted into dRib-1-P (Fig. 2, stage 2), which then reacted with a fluorine-containing base (Fig. 2, stage 3). Stages 2 and 3 were performed in the same flask. To increase the solubility of the base, the reaction was carried out in a buffer solution with the addition of 10 vol % dimethyl sulfoxide.
The concentration of dimethyl sulfoxide in the reaction mixture did not significantly affect the enzymatic activity of PNP, which is consistent with the literature data [18]. The transglycosylation reaction was carried out at different glycosyl donor : base : phosphate ratios (Table 1). Carrying out the reaction with a slight excess of the glycosyl donor in the presence of an equimolar amount of phosphate (1.5 : 1 : 1) or its deficiency (1.5 : 1 : 0.5, 1.5 : 1 : 0.25) led to high yields of target nucleoside products with a slight decrease in the reaction rate compared to the reaction rate in the pres- ence of equimolar amounts of phosphate. An increase in the amount of phosphate in the reaction mixture (starting from an equimolar amount and above) leads to an increase in the rate of formation of dRib-1-P, the hydrolysis of which can reduce the yields of target nucleosides (Table 1, lines [16][17]. Therefore, the highest yields were achieved at a glycosyl donor : base : phosphate ratio of 1.5 : 1 : 0.25 ( Table 2). The yield of the preparative method for obtaining the 5a product was 100% by HPLC (86% after purification by reverse-phase chromatography on silica gel-C18), and the yield of the 5b product was 92% by HPLC (47% after similar purification, the low yield was due to the sorption of the compound on silica gel-C 18 ).
The structure of the obtained compounds was confirmed by NMR spectroscopy. The 13 С-NMR spectrum of the trifluoromethyl-substituted deoxynucleoside 5а contained a resonance signal of the trifluoromethyl group in the form of a low-intensity quartet with a spin-spin coupling constant (SSCC) 1 J C-F = 31 Hz. The 13 С-NMR spectrum of the pentafluorosubstituted deoxynucleoside 5b contained resonance signals of the 13 С nuclei of the phenyl group in the form of three wide doublets with SSCC 1 J C-F of approximately 250 Hz. In the 19 F-NMR spectrum, the signal of the trifluoromethyl group of compound 5а was resolved as a singlet with a chemical shift δ = 61.04 ppm. The 19 F-NMR spectrum of the pentafluoro-substituted nucleoside 5b shows a complex spinspin interaction: a doublet of doublets for the 19 F nuclei in the ortho position of the phenyl substituent with SSCC 3 J F-F = 22 Hz, 4 J F-F = 6 Hz, a triplet of doublets for the 19 F nuclei in the meta position with SSCC 3 J F-F = 22 Hz, 4 J F-F = 6 Hz and a triplet for the 19 F nuclei in the para position of the phenyl ring with SSCC 3 J F-F = 22 Hz. The presence of a pentafluorosubstituted fragment in the 5b structure was also confirmed by the 1 H-NMR spectrum, in which no resonance signals of the phenyl group protons were observed.
The antiviral activity of the obtained compounds is currently being studied.