Construction of an Isonucleoside on a 2,6-Dioxobicyclo[3.2.0]-heptane Skeleton

We have built a new isonucleoside derivative on a 2,6-dioxobicyclo[3.2.0]heptane skeleton as a potential anti-HIV agent. To synthesize the target compound, an acetal-protected dihydroxyacetone was first converted to a 2,3-epoxy-tetrahydrofuran derivative. Introduction of an azide group, followed by the formation of an oxetane ring, gave a pseudosugar derivative with a 2,6-dioxobicyclo[3.2.0]heptane skeleton. The desired isonucleoside was obtained by constructing a purine base moiety on the scaffold, followed by amination.


Introduction
Since the discovery of 3'-azidothymidine (AZT), much attention has been paid to the development of effective chemotherapeutic agents against the human immunodeficiency virus (HIV), a causative agent for AIDS [1,2]. More than 20 anti-HIV drugs have now been approved and are clinically used for the treatment of AIDS. Among them, nucleoside reverse transcriptase inhibitors (NRTIs) play a OPEN ACCESS critical role in the treatment of AIDS patients. In the most successful regimen for AIDS referred to as ART (Anti-Retroviral Therapy), a cocktail of anti-HIV drugs, including NRTIs, non-nucleoside reverse transcriptase inhibitors (NNRTIs), and protease inhibitors (PIs) [3], is used. Although ART greatly contributes to increasing the lifespan of patients, drug-resistant strains of the virus are still a serious problem [4,5]. Therefore, new drugs that are effective against the resistant virus strains are constantly needed.

Results and Discussion
Following our previous reports [11,14], epoxide 7 was synthesized. We first attempted to introduce an adenine onto 7 by treating it with DBU [22]. However, the reaction did not give the desired product 9 (Scheme 1). Scheme 1. Attempt to introduce adenine moiety.
In addition, Lewis acid-catalyzed reactions did not afford 9 either (data not shown). Since the low reactivity of 7 might be due to its rigid structure, we next tried nucleophilic substitution using a more reactive cyclic sulfate derivative [23]. Cis-allyl alcohol 10, a precursor of epoxide 7 [11,14], was cyclized under Mitsunobu conditions, as in the case of epoxide 7 [11,14], to give dihydrofuran 11 in 71% yield. Treatment of dihydrofuran 11 with potassium osmate in the presence of N-methylmorpholine N-oxide afforded cis-diol 12. The desired cyclic sulfate 13 was obtained by treatment of 12 with thionyl chloride, followed by oxidation. However, the nucleophilic substitution of 13 with adenine did not afford the desired isonucleoside 14 (Scheme 2).

Scheme 2.
Second attempt to introduce adenine using a cyclic sulfate. Therefore, we revised our plan to synthesize an isoadenosine constructed on a 2,6-dioxobicyclo[3.2.0]heptane scaffold, and the revised scheme is shown in Scheme 3 in a retrosynthetic manner. Instead of the direct introduction of adenine, we decided to build the adenine ring on the 2,6-dioxobicyclo[3.2.0]heptane pseudosugar skeleton in a stepwise manner. According to this plan, compound 16 was thought to be a suitable intermediate for preparing 15 since it can be transformed to 6 by the formation of an imidazole ring, followed by amination. Fused oxetane derivative 16 can be obtained from dimesylate 17. Finally, epoxide 7, described above, was selected as the starting compound because it can be converted to 17 by the selective cleavage of the oxirane ring with an azide anion (Scheme 2). First, regioselective cleavage of the oxirane ring of 7 with sodium azide in 2-methoxyethanol under reflux conditions gave the desired azide-alcohol 18 as a single regioisomer in 66% yield. It is obvious that the nucleophilic azide anion attacked from the less hindered side since similar regioselective epoxide opening was observed in our previous report [11,14]. After benzoylation of the hydroxyl group, the acetal group of 19 was removed by using acidic hydrolysis, and the resulting diol was mesylated to give dimesylate 20 in good yield. Deprotection of the benzoyl group and the subsequent formation of an oxetane ring were achieved by treating 20 with sodium methoxide under reflux conditions to give mesylate 21 in 72% yield. The structure of 21 was unambiguously determined by comparison of 1D NMR spectrum with that of 2-oxa-6-thiabicyclo[3.2.0]heptane skeleton [14] after converting it to benzoate 22 by treatment with benzoic acid in the presence of cesium fluoride. In 1 H-NMR spectra of 22, the peaks corresponding to the methyl groups of the dimesylate were absent, and only the peaks corresponding to the benzoyl group in the range of 8.1-7.4 ppm were present. In addition, one of the methylene protons at the 2-position was observed as a doublet at 4.42 ppm, meaning that the coupling with H-3 disappeared. This indicates that the conformation around the tetrahydrofuran ring changes and becomes fixed, which causes a loss of coupling between one pair of H-2 and H-3 protons. A similar correlation between conformation and couplings in 1 H-NMR spectra has been reported for the 2-oxa-6-thiabicyclo[3.2.0]heptane skeleton [14]. Moreover, in the mass spectrum of the compound, a molecular ion peak was observed at m/z = 276, further supporting the assignment of the structure. Finally, 22 was deprotected to afford azido-alcohol 23 in 88% yield (Scheme 4).

Scheme 4. Synthesis of isoadenosine 6.
Azido-alcohol 23 was reduced by catalytic hydrogenation to give key intermediate 16, which was treated with 5-amino-4,6-dichloropyrimidine and diisopropylethylamine in refluxing n-butanol [23] to give diaminopyrimidine derivative 15 in 58% yield from 23. Formation of the imidazole ring of 15 was accomplished by treatment with orthoethyl formate under acidic conditions [24] to give 6-chloropurine nucleoside 24. Finally, the isoadenosine was built on the 2,6-dioxobicyclo[3.2.0]heptane scaffold 6 by heating 24 with methanolic ammonia in a sealed tube in 69% yield (Scheme 4). Isoadenosine 6 did not show any significant activity against HIV even at a concentration of 100 µM.

General Information
Melting points are uncorrected. NMR spectra were recorded at 400 MHz ( 1 H), 100 MHz ( 13 C) using CDCl3 as a solvent. As an internal standard, tetramethylsilane was used for CDCl3. Mass spectra were obtained by EI or FAB mode. Silica gel for chromatography was Silica Gel 60N (spherical, neutral, 100-210 µm, Kanto Chemical Co. Inc., Tokyo, Japan). When the reagents sensitive to moisture were used, the reaction was performed under argon atmosphere.

Conclusions
We constructed an isoadenosine derivative on a 2,6-dioxobicyclo[3.2.0]heptane scaffold. Since our initial attempt to synthesize 6 by directly introducing the adenine moiety was not successful, we synthesized it by the de novo synthesis of an adenine ring on a pseudosugar moiety. However, this unique adenosine analogue showed no activity against HIV. Previously, we have reported that neither thymine nor adenine analogues 4 built on a 2-oxa-6-thiabicyclo[3.2.0]heptane skeleton inhibit HIV [14]. The structural rigidities of these analogues and isoadenosine 6 due to the introduction of fused thietane and oxetane rings, respectively, appear to inhibit anti-HIV activity. In particular, phosphorylation at the 5'-hydroxyl group would be inhibited since deoxynucleoside kinase recognizes the puckering of sugars [25]. Thus, we are currently preparing new substituted nucleoside derivatives based on 4 and 6, and the results will be reported elsewhere.