Synthesis of Ribavirin , Tecadenoson , and Cladribine by Enzymatic Transglycosylation

Despite the impressive progress in nucleoside chemistry to date, the synthesis of nucleoside analogues is still a challenge. Chemoenzymatic synthesis has been proven to overcome most of the constraints of conventional nucleoside chemistry. A purine nucleoside phosphorylase from Aeromonas hydrophila (AhPNP) has been used herein to catalyze the synthesis of Ribavirin, Tecadenoson, and Cladribine, by a “one-pot, one-enzyme” transglycosylation, which is the transfer of the carbohydrate moiety from a nucleoside donor to a heterocyclic base. As the sugar donor, 7-methylguanosine iodide and its 2′-deoxy counterpart were synthesized and incubated either with the “purine-like” base or the modified purine of the three selected APIs. Good conversions (49–67%) were achieved in all cases under screening conditions. Following this synthetic scheme, 7-methylguanine arabinoside iodide was also prepared with the purpose to synthesize the antiviral Vidarabine by a novel approach. However, in this case, neither the phosphorolysis of the sugar donor, nor the transglycosylation reaction were observed. This study was enlarged to two other ribonucleosides structurally related to Ribavirin and Tecadenoson, namely, Acadesine, or AICAR, and 2-chloro-N6-cyclopentyladenosine, or CCPA. Only the formation of CCPA was observed (52%). This study paves the way for the development of a new synthesis of the target APIs at a preparative scale. Furthermore, the screening herein reported contributes to the collection of new data about the specific substrate requirements of AhPNP.


Introduction
Nucleoside analogues are well-established drugs in clinical practice; they are mainly used as anticancer and antiviral agents.However, the search for new therapeutically active nucleosides is still a vibrant research area, as witnessed by the approval of the pro-drug Sofosbuvir, marketed as Sovaldi ® in 2013, used in the treatment of hepatitis C as an alternative to peginterferon-combined therapies (e.g., in association with Ribavirin and Daclatasvir, Ledipasvir or Simeprevir) [1].
Drug discovery stands alongside the set-up of new synthetic strategies aimed at circumventing the typical constraints of nucleoside chemistry (e.g., multi-step processes, protection/deprotection reactions, lack of selectivity, etc.) [2].Chemoenzymatic synthesis has been proven to overcome most of these drawbacks.The main advantages of enzymatic methods include high catalytic efficiency, mild reaction conditions (and thus environmentally friendly and safer syntheses), high stereo-and regioselectivity, and fewer numbers of synthetic steps.However, a truly efficient synthesis of nucleoside analogues is often the result of a combination of chemical methods and biochemical transformations [3].
Purine nucleoside phosphorylases (PNPs, EC 2.4.2.1) catalyze the reversible cleavage of the glycosidic bond of purine nucleosides in the presence of inorganic orthophosphate as a co-substrate, to generate the conjugated nucleobase and α-d-pentofuranose-1-phosphate.If a second purine base is in the reaction medium, the formation of a new nucleoside can result by a regio-and stereoselective transglycosylation reaction [3].
Accumulated data about a PNP from Aeromonas hydrophila (AhPNP) [4] have clearly shown that this enzyme can be successfully used in the synthesis of a wide range of nucleoside analogues, which are either routinely used as drugs (e.g., arabinosyladenine) [5,6] or can have promising pharmacological activities, such as some 6-substituted purine ribonucleosides [7,8].This PNP has been shown to have a quite relaxed substrate tolerance toward the purine base, to recognize ribo-and 2 -deoxyribonucleosides as the sugar donor, and, although to a lesser extent, to accept d-arabinose-1-phosphate produced by the phosphorolysis of arabinosyluracil in a bi-enzymatic transglycosylation reaction [4][5][6][7][8][9].
reactions, lack of selectivity, etc.) [2].Chemoenzymatic synthesis has been proven to overcome most of these drawbacks.The main advantages of enzymatic methods include high catalytic efficiency, mild reaction conditions (and thus environmentally friendly and safer syntheses), high stereo-and regioselectivity, and fewer numbers of synthetic steps.However, a truly efficient synthesis of nucleoside analogues is often the result of a combination of chemical methods and biochemical transformations [3].
Purine nucleoside phosphorylases (PNPs, EC 2.4.2.1) catalyze the reversible cleavage of the glycosidic bond of purine nucleosides in the presence of inorganic orthophosphate as a co-substrate, to generate the conjugated nucleobase and α-D-pentofuranose-1-phosphate.If a second purine base is in the reaction medium, the formation of a new nucleoside can result by a regio-and stereoselective transglycosylation reaction [3].
Accumulated data about a PNP from Aeromonas hydrophila (AhPNP) [4] have clearly shown that this enzyme can be successfully used in the synthesis of a wide range of nucleoside analogues, which are either routinely used as drugs (e.g., arabinosyladenine) [5,6] or can have promising pharmacological activities, such as some 6-substituted purine ribonucleosides [7,8].This PNP has been shown to have a quite relaxed substrate tolerance toward the purine base, to recognize ribo-and 2′-deoxyribonucleosides as the sugar donor, and, although to a lesser extent, to accept D-arabinose-1- phosphate produced by the phosphorolysis of arabinosyluracil in a bi-enzymatic transglycosylation reaction [4][5][6][7][8][9].
Ribavirin (Virazole ® ) is considered the "gold-standard" in the treatment of hepatitis C in association with pegylated interferon-alpha (IFN-α) [10][11][12][13].Tecadenoson and its congeners are selective A1 receptor agonists, which have been investigated for their use against arrhythmia and atrial fibrillation [14][15][16].Cladribine (Litak ® ) has been approved for the treatment of symptomatic tricoleukaemia (hairy-cell leukemia).It can act both as a chemotherapy drug and an immunosuppressive agent.Clinical studies have also suggested its potential usefulness in the treatment of multiple sclerosis [17,18].Chemical structures of Ribavirin (1), Tecadenoson (2), and Cladribine (3) are reported in Figure 1.The synthesis of Ribavirin (1) has been achieved by both chemical and enzymatic approaches.The established glycosylation route involves the reaction of peracetylated β-D-ribofuranose with methyl 1,2,4-triazole-3-carboxylate, followed by aminolysis.Despite the good yields (54-83%) as well as the high regio-and stereoselectivity, high temperatures (135-170 °C) and high vacuum (15-55 mmHg) are required for the formation of the glycosydic bond [19,20].Interestingly, when the reaction was carried out under slightly milder conditions (MW irradiation, 130 °C, 5 min and direct MPLC purification), the overall yield of Ribavirin was only 35% [21].
Also, for Cladribine, enzymatic glycosylation (i.e., enzyme-catalyzed formation of the glycosydic bond) has been suggested as an alternative to the chemical route.Some examples of transglycosylation based on the use of PNPs (from E. coli and Geobacillus thermoglucosidasius) in mono-or bi-enzymatic processes have been reported [47,48].
While this research was ongoing, an E. coli PNP-catalyzed transglycosylation for the preparation of 2 -deoxynucleosides, including Cladribine, was developed by Mikhailov and co-workers who exploited a 7-methyl purine nucleoside iodide as the sugar donor [49].
In this context, we explored the use of 7-methylguanine arabinoside iodide (9) (Scheme 1) as the sugar donor for the synthesis of arabinosyl purine analogues such as the antiviral drug Vidarabine, as an alternative scheme to both the conventional chemical synthesis and the bi-enzymatic transglycosylation reaction [6].aminotetrahydrofurane or its salts followed, when necessary, by deprotection with ammonia (reported yield: 68%) [34][35][36].
Also, for Cladribine, enzymatic glycosylation (i.e., enzyme-catalyzed formation of the glycosydic bond) has been suggested as an alternative to the chemical route.Some examples of transglycosylation based on the use of PNPs (from E. coli and Geobacillus thermoglucosidasius) in monoor bi-enzymatic processes have been reported [47,48].
While this research was ongoing, an E. coli PNP-catalyzed transglycosylation for the preparation of 2'-deoxynucleosides, including Cladribine, was developed by Mikhailov and co-workers who exploited a 7-methyl purine nucleoside iodide as the sugar donor [49].
In this context, we explored the use of 7-methylguanine arabinoside iodide (9) (Scheme 1) as the sugar donor for the synthesis of arabinosyl purine analogues such as the antiviral drug Vidarabine, as an alternative scheme to both the conventional chemical synthesis and the bi-enzymatic transglycosylation reaction [6].
The choice of CH3I as the methylating agent and of the proper solvent (DMSO or a DMF/DMSO mixture) led to the selective formation of the iodide salts at N-7 (Scheme 1).Light exposure during the reaction at room temperature was avoided to prevent the decomposition of the final products, which had to be stored at −20 °C until use.The stability of 7-methylated nucleosides in DMSO or DMSO/DMF mixtures was found to depend on the nature of the sugar moiety, and posed a serious issue in the synthesis of 7-methyl-2′-deoxyguanosine iodide (8), as highlighted by 13 C NMR spectra registered in DMSO-d6 (see Supplementary Materials, Figure S1).The methylation reaction was thus performed by modifying a previously reported protocol [50] under controlled temperature (20 °C),
The choice of CH 3 I as the methylating agent and of the proper solvent (DMSO or a DMF/DMSO mixture) led to the selective formation of the iodide salts at N-7 (Scheme 1).Light exposure during the reaction at room temperature was avoided to prevent the decomposition of the final products, which had to be stored at −20 • C until use.The stability of 7-methylated nucleosides in DMSO or DMSO/DMF mixtures was found to depend on the nature of the sugar moiety, and posed a serious issue in the synthesis of 7-methyl-2 -deoxyguanosine iodide (8), as highlighted by 13 C NMR spectra registered in DMSO-d 6 (see Supplementary Materials, Figure S1).The methylation reaction was thus performed by modifying a previously reported protocol [50] under controlled temperature (20 • C), short reaction time, and an excess of CH 3 I in order to avoid any decomposition of 8 in DMSO.Higher temperatures (25-30 • C) and longer reaction times led to product decomposition.
No synthesis of 7-methylguanine arabinoside iodide (9) has been reported in the literature to date; thus, the same strategy (CH 3 I in a DMSO/DMF mixture) was successfully applied to the methylation of arabinosylguanine (6), thus affording 9 in 83% yield.In contrast to 8, 7-methylguanine arabinoside iodide (9) was as stable as the corresponding ribo-derivative (7).No decomposition products were detected both in DMSO/DMF and under bioconversion conditions (see below).

"One-Pot, One-Enzyme" Transglycosylations
"One-pot one-enzyme" transglycosylations were carried out starting from either 7-methylguanosine iodide (7) or 7-methyl-2 -deoxyguanosine iodide (8) as the sugar donors for the synthesis of the APIs 1-2 and 3, respectively (see Scheme 3).As previously reported for nucleoside 7 [4,7], also the phosphorolysis reaction of 8 was almost complete and irreversible.In fact, the conjugated nucleobase of these nucleosides (7-methylguanine, 19, see Scheme 3) was not recognized by AhPNP as a substrate, thus assisting the shift of the reaction equilibrium toward the product formation.
Transglycosylation reactions occur under very mild conditions, generally in phosphate buffer at room temperature.The typical drawback of enzymatic reactions is the need to conjugate the poor solubility of substrates in aqueous media with the stability of the biocatalysts in organic solvent.Starting from a reaction set-up established in our labs, glycerol was used as the co-solvent in order to improve the substrate solubility and to preserve the enzyme activity (glycerol is routinely used as a protein preservative).Only in the case of Cladribine (3), DMSO was added as a second co-solvent besides glycerol (1 mL, 5% of the total volume) to overcome the very poor solubility of 2-chloro-6-aminopurine (11).As previously reported [8], the use of DMSO is quite well tolerated by AhPNP.
As for the solubility issue, 7-methyl purine nucleosides are highly water soluble and their use as the sugar donors is, indeed, a true advantage.On the other hand, the conjugated base of these nucleosides (7-9), i.e., 7-methylguanine (19), is poorly water soluble and easily separates out the reaction, thus giving a further contribution to drive the reaction equilibrium.No less important, a further strength-point of 7-methyl purine nucleosides as the sugar donor relies on their straightforward and high-yielding preparation (see Section 2.1).
All reactions were carried out in 50 mM phosphate buffer (pH 7.5) containing 20% of glycerol (v/v) at room temperature (Scheme 3).Bioconversions were performed at an analytical scale (1 mM substrate concentration) by using a 1:1 donor/acceptor ratio.Reactions were monitored both by measuring the depletion of the nucleobase acceptor as well as the formation of the new nucleoside (see Materials and Methods).Conversions (end-point: 24 h) are reported in Table 1.All the HPLC peaks were assigned on the basis of the pure reference compounds, either purchased or synthesized (see Materials and Methods).As recalled in the Introduction, the synthesis of nucleoside-based APIs by an enzymatic transglycosylation reaction has been investigated by many authors (for a comprehensive review, see References [3,51]; for a recent example see Reference [52]).However, the enzymatic synthesis of Tecadenoson has been reported herein for the first time (see Supplementary Materials, Figure S7).Taking into account this result and the evidence that AhPNP can accept a wide array of 6-substituted purines as substrates, we also successfully synthesized CCPA (17), the congener of Tecadenoson (Scheme 3).In this case, as the base acceptor (12) was not soluble in the buffer-glycerol mixture even in the presence of 5% DMSO (as applied in the synthesis of Cladribine), the biotransformation was carried out in a sort of "fed batch" mode.A 10 mM stock solution of 12 in DMSO was progressively added to the reaction upon monitoring the rate of phosphorolysis and the transglycosylation reaction (see Supplementary Materials, Figure S5).The final percentage of DMSO was 10% v/v.It is worth reporting that a large excess of AhPNP was used in this case, as the formation of the target nucleoside was hardly detectable when using 1.15 U of enzyme.This result prompted us to further extend this approach to the synthesis of other ribo-derivatives such as Acadesine (or AICAR, 18), a congener of Ribavirin.This reaction was performed under standard conditions (glycerol-buffer) by using the same excess of AhPNP as for CCPA.At this stage of the project, in fact, the goal was to assess whether the biocatalyst could synthesize Acadesine.Surprisingly, whereas in the case of Ribavirin a conversion of 67% was registered after 24 h, the formation of Acadesine was not observed (see Supplementary Materials, Figures S4 and S6).
As it is well known, the chemical route to purine 2 -deoxyribonucleosides is even more challenging than that to its ribo-counterparts [2].Therefore, the availability of 7-methyl-2 -deoxyguanosine iodide would represent a valuable tool for an alternative synthetic approach, as proven by the bioconversion of Cladribine (see Table 1).However, this sugar donor was found to be less stable than 7-methylguanosine iodide in aqueous medium, in agreement with a very recent report by Mikhailov et al. [49].We have found that this molecule is unstable also in DMSO at room temperature, as indicated from the appearance of extra signals in the 13 C NMR spectrum recorded at two-hour intervals in DMSO, clearly showing that the decomposition of the nucleoside had occurred (see Supplementary Materials, Figure S1).Dimethyl sulfoxide was used in the synthesis of Cladribine (3) as the second co-solvent.The evidence of the poor stability of 7-methyl-2 -deoxyguanosine iodide both in buffer solutions and in DMSO makes this molecule a substrate which is difficult to handle in preparative applications.
As a natural continuation of this study, our efforts were then focused on the synthesis of arabinosyl purine nucleosides by using the newly prepared 7-methylguanine arabinoside iodide (9) as the sugar donor.Vidarabine (arabinosyladenine) was selected as the target API as its enzymatic synthesis, although through a bi-enzymatic approach [6,9], it was successfully achieved even at a preparative scale with a good yield and purity (3.5 g/L, 53% yield, 98.7% purity).Surprisingly, when using 7-methylguanine arabinoside iodide (9), neither phosphorolysis, nor transglycosylation, thereof, were observed.This result suggests the need for a deeper understanding of the structural requisites for the enzyme-substrate molecular recognition; this is a necessary step to rationalize all the data collected over the years about the substrate specificity of AhPNP as well as to further widen the exploitation of this enzyme in the bio-catalyzed synthesis of modified nucleosides, also at a preparative scale.In this regard, taking into account that the synthesis of Ribavirin (1), Tecadenoson (2), and Cladribine (3) were performed under screening conditions by using a 1:1 donor/acceptor ratio, conversions values were remarkable (≥50%) and foresee considerable room for improvement.
1 H and 13 C spectra were recorded at 400.13 and 100.61Hz, respectively, on a Bruker AVANCE 400 spectrometer equipped with a TOPSPIN software package (Bruker, Karlsruhe, Germany) at 300 K, unless stated otherwise. 1H and 13 C chemical shifts (δ) are given in parts per million and were referenced to the solvent signals (δ H 3.31-δ C 49.00, δ H 2.50-δ C 39.52 ppm from tetramethylsilane (TMS) for CD 3 OD and DMSO-d 6 , respectively).The 13 C NMR signal multiplicities were based on APT (attached proton test) spectra.The 13 C NMR signals were assigned with the aid of 1 H- 13 C correlation experiments (heteronuclear multiple quantum correlation spectroscopy, HMQC, and heteronuclear multiple bond correlation spectroscopy, HMBC).Electrospray ionization mass spectra (ESI-MS) were recorded on a ThermoFinnigan LCQ Advantage spectrometer (Hemel Hempstead, Hertfordshire, UK).
The pH measurements were performed by using a 718 Stat Titrino pHmeter from Metrohm (Herisau, Switzerland).
Enzymatic reactions were monitored by HPLC using a Merck Hitachi L-7000 La-Chrom liquid chromatographer equipped with a UV-Vis detector, an autosampler (injection volume: 20 µL), and a column oven (instrument 1), or a Chromaster 600 bar system, Merck Hitachi VWR equipped with a UV-Vis detector, an autosampler (injection volume: 20 µL), and a column oven (instrument 2).

Table 1 .
Synthesis of the target nucleosides 1