The selection of effective enzymes for the synthesis of ester prodrugs from racemic 9-(2,3-dihydroxypropyl)adenine in dimethyl formamide (Scheme 1) was performed following a well established screening procedure. The active and selective enzymes, lipases from Pseudomonas fluorescens (free form), Candida antarctica (free form), Aspergillus niger (free form), hog pancreas (free form) and Geotrichum candidum 4013 (cell-bound lipase as the acetone powder, immobilized cell-bound lipase and immobilized extracellular lipase) were used for preparative scale purposes.

It was previously described that lipase from Aspergillus oryzae had a higher specificity toward triacylglycerols of medium-chain saturated fatty acids over short-chain or long-chain fatty acids. The enzyme demonstrated 1,3-positional specificity [13]. Berger and Schneider [14,15] developed a facile system for the quantitative determination of lipase regioselectivities in organic solvents towards the 1,3 position of glycerides. Their study showed that the lipases from Mucor meihei and Pseudomonas fluorescens did not display high 1,3-positional specificities. It was reported that the broad isoelectric region of Candida antarctica lipase B is unique as compared to almost all other α/β-hydrolases which have a well-defined isoelectric point [16]. However, for some lipases the chain length of the fatty acid moiety of the ester substrate can have an effect on reaction rate. It was reported that the lipases from C. rugosa and porcine pancreas exhibited a preference for shorter-chain substrates, whereas the lipase from R. miehei showed little significant difference in transesterification reaction rate with a range of C₈–C₁₈ fatty acid moieties [17].

The lipases from Geotrichum candidum were used as a biocatalyst in the transesterification of DHPA as well. It was already found that the fungus Geotrichum candidum 4013 produces two types of lipases (extracellular and cell-bound [7]). The extracellular lipase displays low selectivity against polyunsaturated fatty acids, while the cell-bound lipase possesses selectivity to saturated fatty acids [18]. Enantiomeric ratio (E) achieved with both induced enzymes in the tested processes was high (from 43 to 242, [19]). Until now the catalytic activities of the enzymes were tested using acetone powder (cell-bound lipase [20]) or lyophilized powder (extracellular lipase, [20]) and they exhibited excellent enantioselectivity and chemical yields. In the present study we decided to use the enzymes in their immobilized form, because an adequate immobilization procedure can greatly improve and modulate the catalytic properties of lipases like thermostability and selectivity [21]. Another benefit of using immobilized enzyme is reusability of the biocatalysts. Therefore lipases from G. candidum 4013 (extracellular and cell-bound) were immobilized to chitosan beads. Chitosan exhibits several advantages as an immobilization carrier, including low biodegradability and easy handling.

Since the amount of enzyme bound to the support, as well as their activity, would be largely influenced by the time and temperature of immobilization, these parameters were optimized initially (data not shown). An immobilization time of 3 h was found to be optimal at 25 °C. Under these conditions, a protein loading from 33% to 48% was obtained. The hydrolytic activity of the immobilized extracellular lipase was 0.052 U/g chitosan beads (Table 1) and the activity of immobilized cell-bound lipase was 2.5 higher in comparison to the activity of immobilized extracellular lipase.

It is known that the nature of the solvent and the type of nucleophile can significantly influence the enzymatic esterification. Herein dimethylformamide has been chosen as the best solvent because of the low solubility of DHPA in the other commercially available solvents. The effect of chain length of the fatty acid moiety of vinyl esters on lipase activity was tested. Generally, the chemical yield of the reaction decreased when acyl donors with a higher chain length were used (Table 2). The only exception was observed when lipase from P. fluorescens was used as biocatalyst in transesterification between vinyl butyrate and DHPA. Under these conditions the highest chemical yield among all obtained results was 72%. In the case of vinyl palmitate all of the tested enzymes were catalytically inactive.

Sample composition was determined by HPLC performed on a Waters 600E system. By this analytical method any remaining DHPA leaves the column first, with a retention time of 5.1 min, followed by monoacylated compounds with retention times ranging from 7 to 16 min (2a = 7.2 min, 2b = 10.2, 2c = s 13.7 min and 2d = 15.7 min) at 9.3 ± 0.2 MPa. Diacylated compounds are supposed to be formed, but we determined only monoacylated mixtures of two such products. Diols, which contain secondary and primary hydroxy groups, can be selectively monoacylated at the primary hydroxy group by many lipases in organic solvents. Since the reaction does not take place at the chiral secondary center itself, observed enantioselectivities are usually low or enantioselectivity was not observed at all [22]. Usually this kind of reaction proceeds with very low (or none) enantioselectivity, although some specific classes of primary alcohol can be acylated with high enantioselectivity by mean of lipase catalysis [23,24,25]. The phenomenon that diols are inferior substrates than their corresponding monoacylated derivatives in lipase catalyzed transesterification is generally known. It was also observed for racemic diols with C2 symmetry [26,27] where the first acylation step is less enantioselective than the second one. An analogous behavior was found for many prochiral diols [28,29,30,31] in which the monoacylation step shows a low or only moderate enantioselectivity, but in many cases the low enantioselectivity can be overcome by application of a sequential lipase-catalyzed acylation procedure [22].

Satisfactory chemical yields were obtained with lipase from Candida antarctica and lipase from Pseudomonas fluorescens when they were used as biocatalysts (Table 2). In the case where acetone powder of Geotrichum candidum cells was used as biocatalyst the chemical yield of the transesterification was comparable with results obtained using commercially available enzymes. On the other hand, both lipases from Geotrichum candidum 4013 immobilized on chitosan beads provided less catalytic activities than the free forms. Generally it is known that the immobilization of enzymes has an effect on the catalytic properties of enzyme. Various methods of immobilization of lipases on many different types of supports are available. Our experimental investigations have produced unexpected results, such as a significant reduction in lipase activity compared with the acetone powder form of the enzyme. A more detailed search for a convenient support for the immobilization of lipases will be the most critical point for resolving this problem. Other key components the type of support used for immobilization, conjugation chemistry, conjugation conditions, and selection of reaction a medium all have complex effects on the enzyme activity for a particular substrate and will be optimized accordingly.

One of the important characteristics of an immobilized enzyme is its stability and reusability over an extended period of time. The reusability of both the immobilized lipase from G. candidum was evaluated in the transesterification of DHPA (see Experimental section) at 25 °C and 70% of the initial activity was retained after 2 runs. Between cycles, the catalysts were washed with dry hexane and DMF and thereafter used without drying. Our results coincide with results previously published [32,33]. The repeated use of the immobilized P. chrysogenum lipase in the hydrolysis of olive oil was studied. After six reuses, immobilized lipase retained 72.09% of its activity [34]. Hung and coworkers found that with repeated use immobilized C. rugosa lipase on chitosan retained 74% after 10 reuses [32]. Similarly Yi and coworkers showed that the activity of immobilized lipase of C. rugosa on alanine chitosan beads retained 77% of the initial activity after 10 reuses [35]. Antiviral activities of all the above described DHPA prodrugs is currently being investigated by the team of Professor Jan Balzarini at the Rega Institute for Medical Research KU Leuven (Belgium).

The strain of Geotrichum candidum 4013 was obtained from the Culture Collection of the Department of Biochemistry and Microbiology (DBM), Institute of Chemical Technology, Prague. Lipases [hog pancreas (2.4 U/mg), Aspergillus niger (4 U/g), Candida antarctica (3.0 U/mg) Pseudomonas fluorescens (42.5 U/mg)] and chemicals were purchased from Sigma Aldrich (St. Louis, MO, USA). Racemic 9-(2,3-dihydroxypropyl)adenine (DHPA) was prepared in the laboratory of Dr. Krečmerová.

For solid culture, Geotrichum candidum 4013 was inoculated from culture slant. Growth medium with the following composition per liter was used: glucose (30 g), corn steep (10 g), MgSO₄·7H₂O (0.5 g), KH₂PO₄(1 g), K₂HPO₄(1 g), NaNO₃(2 g), KCl (0.5 g), FeSO₄·7H₂O (0.02 g) and agar (15 g). The tubes were incubated at 25°C for 3 days and conserved at 4 °C. Liquid culture was prepared with medium consisting of glucose (30 g), corn steep (10 g), MgSO₄·7H₂O (0.5 g), KH₂PO₄(1 g), K₂HPO₄(1 g), NaNO₃(2 g), KCl (0.5 g), FeSO₄·7H₂O (0.02 g) in a simple conical flask (250 mL) containing 100 mL of the medium and closed with sterile stopcocks. The medium was sterilized at 121 °C for 20 min. The medium was inoculated with the cells originally growing on culture slants and incubated with shaking at 30 °C for 24 h. For the activation of the lipase production (extracellular and cell-bound) the medium containing per litre: peptone (50 g), glucose (10 g), MgSO₄·7H₂O (1 g), NaNO₃(1 g), olive oil (10 g) was used. Medium (90 mL) was inoculated with 10 mL of prepared inoculum. After cultivation the broth was filtered. Cell pellet was used as source of cell-bound lipase and supernatant as crude extracellular lipase.

Cells were harvested by filtration of 100 mL cell suspension (inoculated medium) through a 0.2 µm filter, the remaining eluate was washed in distilled water (100 mL). The cells of G. candidum 4013 (3 g wet wt) were mixed with cold acetone (−20 °C, 100 mL), and the cells were collected by filtration. The procedure was repeated five times, and then the cells were dried under atmospheric pressure and ambient temperature. The dried cells (0.5 g) were obtained and used without further purification as crude enzyme (acetone powder) and refrigerated under nitrogen for storage.

The cell pellet was separated from the broth (100 mL) by filtration and washed by ice-chilled acetone (200 mL). The suspension was filtrated and the pellet was dissolved in 200 ml of acetone (−20 °C). The suspension was vigorously stirred 2 h. After filtration of the mixture, the pellet was dissolved in 200 ml of diethyl ether (−20 °C), filtered and dried at room temperature on the filter. Enzyme was extracted from the cells by 0.05 M Tris HCl buffer pH 8.0 (5 mL) at room temperature with constant stirring for 90 min. The cell free extract was obtained by filtration. The crude extract containing free lipase in the aqueous phase was stored at 4 °C.

Preparation of chitosan beads: 3% (w/v) chitosan powder was dissolved in 1% acetic acid. Spherical beads of diameter in the range 1–2 mm were produced by adding the chitosan solution dropwise into a coagulant bath consisting of 1M NaOH containing 26% (v/v) ethanol under stirring. Allowing the mixture to remain overnight, the spherical beads were removed by filtration and washed with deionized water until neutrality. The beads then were stored in deionized water at 4 °C until use. Chitosan beads (18 g) had been previously soaked in hexane under agitation for 1 h. Then, excess hexane has removed followed by the addition of 60 mL of crude extracelular lipase (supernatant, concentration 1.86 mg protein/mL) or 60 mL of the crude extract containing free lipase in the aqueous phase (4.70 mg protein/mL). The mixture was agitated for 3 h at room temperature followed by an additional period of 18 h under static conditions at 4 °C. The immobilized lipase on chitosan beats was filtered off, rinsed with hexane and was stored in water at 4 °C until use.

The lipase activity of all used enzymes was determined before our experiments by a spectophotometric method (Table 3). The activity assays of commercial enzymes declared by the supplier were different concerning substrate, pH and temperature.

The release of yellow p-nitrophenol due to hydrolysis of p-nitrophenyl laurate by lipase was measured. A reaction mixture containing 3 mM p-nitrophenyl laurate (100 µL, dissolved in 2-propanol), 50 mM Tris-HCl (1 mL, pH 7.5), and 50 µL of crude extracellular or free cell-bound lipase or 50 mg of cell-bound lipase as acetone powder or fresh cells (prepared as mentioned above) was incubated at 25 °C. Since autohydrolysis of substrates produced low but significant background values at 410 nm, the absorbance in each assay was measured against a substrate-buffer mixture. The p-nitrophenol released was monitored spectrophotometrically at 410 nm. One lipase unit (U) was defined as the amount of enzyme that released 1 µmol p-nitrophenol per minute.

An enzyme (0.025 U, commercially available lipases or immobilized lipases prepared according to the paragraph above) was added to a solution of the substrate DHPA (22 mg; 0.1 mmol) in DMF (1.5 mL). The reaction was started by addition of vinyl esters (0.5 mL or 0.2 mmol of vinyl palmitate in 0.5 mL of DMF). The reaction was performed in flasks under stirring at 25 °C for more than 6 days. The progress of the reaction was monitored by TLC analysis. Final work-up consisted of filtration off of the enzyme and evaporation of the solvent, and chromatographic separation of the residue.

The ¹H-NMR and ¹³C-NMR spectra were recorded in DMSO–d₆ on Bruker Avance II 600 and/or Bruker Avance II 500 spectrometers operating at 600.0 or 500.0 MHz in ¹H and 150.9 or 125.7 MHz in ¹³C. TLC was carried out on precoated silica gel TLC plates. A column (250 mm × 4 mm) filled with a Separon SGX C₁₈ solid phase (5 μm; Watrex) was employed for HPLC analysis of sample composition using gradient of Solvent A (water) and Solvent B (acetonitrile): From 100 to 0% Solvent A in the 30 min as mobile phase at 1 mL min⁻¹. Detection of the compounds during the HPLC analysis was at 254 nm. Products of the reactions were characterized by ¹H-NMR and ¹³C-NMR analysis. Optical rotation was measured on Perkin-Elmer 241 polarimeter and mass spectrum was obtained from LTQ Orbitrap XL (Thermo Fisher Scientofic).