Syntheses of Isoxazoline-Carbocyclic Nucleosides and Their Antiviral Evaluation: A Standard Protocol

The current synthesis of racemic purine and pyrimidine isoxazoline-carbocyclic nucleosides is reported, detailing the key-steps for standard and reliable preparations. Improved yields were obtained by the proper tuning of the single synthetic steps, opening the way for the preparation of a variety of novel compounds. Some of the obtained compounds were also evaluated against a wide variety of DNA and RNA viruses including HIV. No specific antiviral activity was observed in the cases at hand. Novel compounds were prepared for future biological tests.


Introduction
Carbocyclic nucleosides (carbanucleosides) [1] are modified nucleosides [2] in which a methylene group has replaced the oxygen atom of the furanose ring [3]. Due to the absence of a glycosidic linkage, carbanucleosides are chemically more stable and not subjected to the phosphorylases that cleave the N-glycosidic linkage in conventional nucleosides [4]. Many compounds of this class with potent and selective biological activity have been identified [5], including carbovir [6] 1 and abacavir [7] 2, which are currently used therapeutically as anti-HIV agents.
An important aspect of the anti-HIV therapy is the suppression of viral replication in the brain; in this regard an enhanced lipophilicity of potential drugs is likely to be advantageous. Recently, Vince and Hua [6] have reported the modification of the stavudine 3 by fusion of a benzene ring to the carbasugar, leading to analogue 4 [8]. This structural modification increases the lipophilicity (and hence the access to the central nervous system, a major reservoir of HIV) [9] while maintaining the rigidity imposed by the C2 -C3 double bond. Analogously, carbanucleosides in which an aromatic heterocycle is fused to the carbasugar of carbovir and abacavir have been prepared [10]. Five-membered heterocyclic rings such as a pyrazole [11] or isoxazoline [12] are often found in carbocyclic nucleosides [13] as expedients to tune up the lipophilicity or reduce the conformational flexibility of the carbocyclic moiety ( Figure 1) [14].
Recently, we have reported the synthesis of carbocyclic nucleosides 5 containing a fused isoxazoline ring and lacking a methylene (CH 2 ) group in the side chain in the carbocyclic unit [12][13][14]. Nucleosides lacking a methylene group in the side chain have been reported and in some cases display reduced cytotoxicity [15]. The synthetic route that we have exploited, based on the hetero-Diels-Alder (HDA) cycloaddition of dienes with transient acyl nitroso moieties [16], affords "privileged structures" [17] for the synthesis of carbocyclic nucleosides [18].
However, some steps of the synthetic procedure for the linear construction of the nucleobases were problematic and not all of the synthesized compounds could be prepared on a suitable scale for biological tests. Because of these problems, at the beginning of our work, it was not possible to fully evaluate the biological activity of this class of nucleosides, and further structural modifications, necessary for an improved antiviral activity and for a better understanding of the structure-activity relationship (SAR), were sometimes forbidden. On pursuing our studies on the biological activity of carbanucleosides, we present here the improved synthetic procedure of compounds of type 5 ( = 1), amenable to a larger scale preparation and for the preparation of a variety of novel compounds. Their inhibitory activity against a wide range of DNA and RNA viruses, including HIV, is also reported and discussed.

Results and Discussion
2.1. Chemistry. The HDA cycloadduct N-benzoyl-2,3oxazanorborn-5-ene 8 was obtained from freshly distilled cyclopentadiene 6 (2 equivalents) that quantitatively trapped the nitrosocarbonyl benzene 7 generated in situ through the mild oxidation of benzonitrile oxide Ph-CNO (BNO) in CH 2 Cl 2 with a slight excess (1.3 equivalents) of N-methylmorpholine N-oxide (NMO) according to the published procedure (Scheme 1) [18,19] in good yields (73%). It is essential not to exceed with the NMO amounts in this preparation because a larger excess of the oxidant could be detrimental for the nitrosocarbonyl intermediate's life, since the N-oxide could add the intermediate and promote its decomposition [16,18,19]. In Scheme 1 the decomposition pathway is sketched. Nitrosocarbonyls are in fact short-living species, just 180 s as reported by Cohen and coworkers [20] and, if not instantly trapped, can add the excess NMO to give the intermediate 7i that evolves to add nucleophilic species which can be present in the reaction mixtures even in small amounts to afford N-methylmorfoline, N 2 O, and the adduct between the benzoyl group and the nucleophile Nu (in case of water, benzoic acid) [21].
The N-benzoyl-2,3-oxazanorborn-5-ene 8 was found to be an excellent dipolarophile towards BNO addition, affording the two regioisomeric cycloadducts 9a and 9b which were easily isolated in quantitative yields by chromatographic separation in a 3 : 2 regioisomeric ratio [18,19]. A proper tuning of the subsequent synthetic steps had to be done for the scale-up necessary for the synthesis of nucleosides in the due amount for the biological tests.
The ease detachment of the benzoyl group from the cycloadducts 9a,b was achieved through NaOH/MeOH treatment at room temperature for 24 h, affording the fairly stable hydroxylamines 10a,b. The careful revision of the alkaline hydrolysis allowed for improved protocols for this step that was forced to take place in quantitative yields (+20-30% with respect to the previous method) [19]. The reaction has been conducted by adding a slight excess (1.2 equivalents) of NaOH, added portionwise in 2-3 h, to a methanol solution of cycloadducts 9a,b, stirred at room temperature until complete disappearance of the starting materials. Extended reaction times (from 12 to 24 h) ensured the quantitative transformation of the HDA cycloadducts 9a,b into the desired hydroxylamines 10a,b. These compounds were previously fully characterized [19] and found completely stable in the solid phase but they do not survive for long when left in solution and decompose in several uncharacterized products in miserable yields.
The hydrogenolytic cleavage of the N-O bond of the hydroxylamine 10a,b has been also improved and takes place also quantitatively (+30-40% with respect to the previous method) [19] by performing the reaction with Pd/C 10% in ethyl acetate at room temperature for 3 h (extended time). Hence, the aminols 11a,b were obtained and found identical with authentic samples previously prepared [19]. The regioisomeric aminols 11a,b were then submitted to the transformations into the desired nor-nucleosides [13] through linear construction of the purine and pyrimidine rings [22].
The synthesis of the purine-nucleosides 13 requires the two steps sketched in the Scheme 2.
The first one is the condensation of the regioisomeric aminols 11a,b with the 5-amino-4,6-dichloropyrimidine to give the pyrimidine derivatives of type 12a,b. This condensation is indeed the most difficult step in the synthesis, which attained only low yields (around 50%) in the previously reported preparations [13]. Better yields of 12a,b were achieved by increasing the reaction temperature gradually up to the boiling point of the solvent. An excess (2 equivalents) of the 5-amino-4,6-dichloropyrimidine was added to a n-BuOH solution of the aminols 11a,b in the presence of 5 equivalents of i-Pr 2 EtN. The solutions are stirred at room temperature for 1 h and a gradual increase of temperature at the boiling point of n-BuOH (117 ∘ C) was conducted in 1 h in sealed tubes. After 48 h under these conditions the n-BuOH was removed at reduced pressure and the residues were taken up with dichloromethane (DCM) and washed with water. The residues, obtained by evaporation of the DCM, were submitted to chromatographic separation to isolate, from the excess of reagent, the pyrimidine derivatives 12a,b in 79% yields for both regioisomeric compounds. Any attempt to shorten the reaction time by using microwave irradiation instead of conventional heating unfortunately failed. The structures of compounds 12a,b were confirmed through comparison with authentic samples and spectroscopic analyses. For NMR details we send back to [13].
The conversion of the pyrimidine derivatives 12a,b into the chloropurine compounds 13a,b, which was found previously somewhat problematic due to incomplete ring closure and instability of final products [13], has been also simplified.
Solutions of 12a,b in ethyl orthoformate were stirred at room temperature until the pyrimidine derivatives were completely converted into the formylated compounds (monitoring by TLC, 5-7 h) and at this time a catalytic amount of pTsOH was added and stirring continued for 5 days. On applying the frequently reported methods [23] using triethyl orthoformate in the presence of 37% HCl at r.t., no condensation took place and the starting materials were recovered unchanged after the suggested work-up. Harsher conditions [24] sometimes allow for the obtaining of the desired product albeit in poor yields, but in most of the cases the starting material decomposed. The rather tedious work-up procedure (orthoformate evaporation, addition of Et 3 N to a CHCl 3 solution and stirring for 24 h, washing with water and drying, evaporation of the CHCl 3 , residues taken up with EtOAc and stirring for several hours, evaporation of the solvent, and chromatographic separation) has been simplified by adding water to the orthoformate solutions  to hydrolyze the excess orthoformate and the formylated intermediates and allowing the solution to stand at room temperature for 24 h. Extraction with DCM allowed isolating in quantitative yields (+10%) the chloropurine derivatives 13a,b which have been characterized spectroscopically and found identical with authentic samples (see [13] for NMR details). The final chlorine replacement method with selected amines did not require any revision since the substitution occurs with high yields and was conducted as previously described [13] upon heating MeOH solutions of 13a,b at 50 ∘ C in the presence of an excess NH 3 or some primary amines (methyl-, benzyl-, and cyclopropyl-amine) affording the adenine derivatives 14a,b in more than 92% yields (HPLC) which have been submitted to the biological tests ( Figure 2).
However, this type of derivatization is not applicable to all the groups to be inserted as nucleophiles. Alcoholic solvents remain the base to dissolve properly the chlorine-nucleosides but the presence of a cosolvent has to be evaluated, accordingly [25]. This is the case of the insertion of a thio-group (SH) in place of the chlorine atom (Scheme 3).
The chlorine-nucleosides 13a,b were dissolved in an hydroalcoholic solution in the presence of excess Na 2 S and the mixture was heated at reflux for a couple of hours. Upon neutralization with acetic acid 10% and cooling, the thio-derivatives 14a,bS crystallized and were easily filtrated (14a,bS, 89% and 87% yields, resp.). Their structures rely upon the corresponding analytical and spectroscopic data and this characterization evidenced the presence of the tautomeric equilibrium between the two forms reported in Scheme 3.
The tautomeric ratio varied upon the polarity of the solvent and Table 1 reports the ratios determined in representative polar solvents and the chemical shifts of the main protons of the structures reported. The ratios were determined through NMR spectroscopic studies by measuring the integral ratios of the CH=N sharp singlets in the solvents reported and the attribution to the two tautomers of the different groups of signals was done upon COSY experiments.
Upon comparison of the NMR spectroscopic data of compounds 14a,bS and the NMR data reported in literature [26], it was possible to attribute unequivocally the structures of the two tautomeric forms for both the regioisomers [27]. It is known in fact that the thioxo form II is greatly stabilized in aqueous solution [28]. The NMR data in our hands suggest that the form II is the major component in some of the mixtures and the ratios can be reverted by changing the solvent. The equilibrium is slow and, besides the aromatic signals of the phenyl groups, which are coincident, most of the other signals both from the purine heterocyclic ring and the isoxazoline-cyclopentanol moiety differ remarkably allowing for the clear resolution of the two tautomers.
A different behaviour is noticeable between the two regioisomers 14aS and 14bS. In deuterated acetone, regioisomer 14aS is mainly present as the thioxo tautomer II whose proton signals are downfield shifted with respect to tautomer I in nice keeping with the observations reported in literature for thiopurine derivatives [28]. The two tautomers equally coexist in deuterated methanol while in DMSO the ratio is remarkably reverted in favour of the mercapto form I. This is presumably due to the known ability of DMSO to establish 6 The Scientific World Journal H-bonds, in this case with the SH group, hence stabilizing the tautomer I. The second regioisomer 14bS shows a diminished variation of the tautomeric ratio as a function of the solvent changes. Both in DMSO and deuterated methanol the ratios are close to 1/1. Only in acetone tautomer I is twice the other.
The stereoisomeric aminols 11a,b were also converted into the uracil and thymine nucleosides. Among the several methods reported in literature for the synthesis of pyrimidines [29] and in particular those involving the formation of two bonds suitable for uracils preparation, the linear construction of these heterocycles [30] through condensation of aminols 11a,b with the appropriate isocyanates was preferred because of its wide applications in a variety of substrates and experimental conditions as well as for the excellent results achieved [31].
The synthetic route to uracil and thymine nucleosides involves the steps illustrated in Scheme 4 and started with the preliminary preparation of the isocyanates whose quality and purity strongly determine the yields of the final compounds. The condensations of the isocyanates with aminols 11a,b were performed at -20 ∘ C in anhydrous DMF solutions and the reactions left at room temperature for 12 h. The urea adducts 15a,b were obtained in 82% yields and found identical with authentic samples previously prepared [13]. Cyclization of the ureas 15a,b took place smoothly upon gentle refluxing in 2 M H 2 SO 4 solution for 3 h. The uracil nucleosides 16Ua,b and the thymine analogues 16Ta,b were isolated from these solutions after pH adjustment to 7 and extraction with dichloromethane. The yields of the cyclization steps were satisfactorily improved (95%, +20%) and the structures confirmed through spectroscopic analyses.  The products were submitted to biological evaluation ( Figure 2). Activity. Compounds 14a,b and 16a The antiviral activity of the above-reported compounds was tested in vitro in HEL, HeLa, and Vero cells cultures, where appropriate, along with the reference antiviral compounds such as brivudine, ribavirin, acyclovir, ganciclovir, and (S)-DHPA. Results are, respectively, shown in Tables 2-4. All synthesized compounds showed no specific antiviral effects; some differences between the regioisomeric nucleosides can be however noted. Regioisomers of type b, involving a phenyl group distal to the heterocyclic base, display better response than regioisomers of type a, which have proximal phenyl and heterocyclic base. If this indication will be confirmed in other cases, future syntheses will benefit of new efforts to orientate more regioselectively the preparation of new nucleosidic targets. Moreover, all compounds were evaluated for antiviral activity against HIV-1 (strain III B ) and HIV-2 (strain Rob) in MT-4 cell culture and none of these compounds showed inhibitory activity at concentration up to 400 g/mL.

Antiviral
The modest antiviral activity showed from these compounds could be likely linked to the lack of substrate activity for cellular and/or viral nucleoside kinases or alternatively the lack of recognition of the compounds by the viral DNA or RNA polymerases.

Conclusions
In conclusion, the synthesis of isoxazoline-carbocyclic nucleosides was properly tuned as well as improved and a variety of analogues were attained starting from the stereodefined heterocyclic aminols 11a,b. These latter ones are readily available through exo selective 1,3-dipolar cycloadditions of benzonitrile oxide to N-benzoyl-oxazanorbornene 8 and simple elaborations of the cycloadducts 9a,b. The stereodefined heterocyclic aminols 11a,b afford the carbocyclic skeleton for the linear construction of the purine type, uracil, and thymine moieties.
Compounds 14a,b and 16a,b were evaluated for their inhibitory activity against a wide variety of viruses through in vitro tests in HEL, HeLa, and Vero cells [32]. Modest antiviral activities were observed. In order to have more insights on the SAR [33] and to increase the antiviral activities of this class of nor-nucleosides their transformation into the corresponding nucleotide monophosphate analogue by esterification of secondary hydroxyl group into phosphonomethyl ether (=P (OH) 2 -CH 2 -O-), which is isosteric of phosphate group (=P (OH) 2 -O-CH 2 -), will be further performed [34].
The biological tests of the thio-derivatives are currently under evaluation. In view of the results obtained with the amino-derivatives, different targets for both the typologies of compounds have been chosen and the results will be accounted in future papers.

Experimental
All melting points are uncorrected. IR spectra (Nujol mulls) were recorded on a FT-IR Perkin-Elmer RX-1. 1 H-and 13 C-NMR spectra were recorded on a Bruker AVANCE 300 in the specified deuterated solvents. Chemical shifts are Materials. Regioisomeric aminols 11a,b were prepared through previously reported syntheses [19] by applying the modified protocols described in this paper to get the aminols in quantitative yields from the regioisomeric cycloadducts 9a,b. Derivatives 12a,b. To aminols 11a,b (3 g, 13.7 mmol) dissolved in n-BuOH (75 mL) and

Synthesis of the Pyrimidine
i-Pr 2 NEt (8.88 g, 68.7 mmol), 5-amino-4,6-dichloropyrimidine (4.50 g, 27.5 mmol) was added. The mixtures were stirred at room temperature for 1 h and then slowly heated at 117 ∘ C with stirring for 48 h. The cooled solutions were evaporated to dryness, taken up in CH 2 Cl 2 , washed with brine, and dried over anhydrous Na 2 SO 4 . The crude residues were then submitted to column chromatography to separate the excess of aminopyrimidine from adducts 12a,b which were isolated in 79% yield for both regioisomeric compounds. Yields improvements are around 30% with respect to previous method. The analytical and spectroscopic data are in full accordance with those reported in [13]. Nucleosides 13a,b. A solution of pyrimidine derivatives 12a,b (3.50 g, 9.8 mmol) in triethyl orthoformate (30 mL) is stirred at room temperature until the starting material has been completely formulated (TLC monitoring) usually in 3-4 h. A catalytic amount of pTsOH was added afterwards and the reaction was stirred at room temperature for 5 days. After this period of time, the excess orthoformate was hydrolyzed in 24 h by adding water and the  water phase extracted with CH 2 Cl 2 and dried over anhydrous Na 2 SO 4 . After evaporation of the solvent, the residue was taken up with diisopropyl ether and few drops of ethyl acetate and the chloropurine derivatives 13a,b crystallized. Yields improvements are around 10% with respect to previous method. The analytical and spectroscopic data are in full accordance with those previously reported [13].  14a,b(A-D). Solutions of chloro-nucleosides 13a,b (50 mg, 0.14 mmol) in MeOH (3 mL) were saturated with ammonia or other gaseous amines and kept in a sealed tube at 50 ∘ C for 24 h. In the case of liquid amines, an excess (40 equiv.) was added to the solutions. The solutions are then cooled and concentration of the solutions afforded oily residues from which the amino