COMPARATIVE STRUCTURAL STUDIES OF [3.1.0]-FUSED 2',3'-MODIFIED BETA-D-NUCLEOSIDES BY X-RAY CRYSTALLOGRAPHY, NMR-SPECTROSCOPY, AND MOLECULAR MECHANICS CALCULATIONS

A structural study is reported on the [3.1.0]-fused nucleosides 2',3'-dideoxy-2',3'-alpha-methyleneuridine (1), 1-(2',3'-dideoxy-2',3'-epimino-beta-D-ribofuranosyl)uracil (2), 1-(2',3'-dideoxy-2',3'-epithio-beta-D-ribofuranosyl)uracil (3), 2',3'-O-anhydroadenosine (4), 1-(2',3'-dideoxy-2',3'-epithio-beta-D-lyxofuranosyl)uracil (5), 1-(2',3'-O-anhydro-beta-D-lyxofuranosyl)uracil (6), 9-(2',3'-O-anhydro-beta-D-lyxofuranosyl)adenine (7), and 1-(2',3'-O-anhydro-beta-D-lyxofuranosyl)thymine (8). Note that compounds 1-4 have the three-membered fused ring in the exo orientation (alpha-face) and compounds 5-8 have the three-membered fused ring in the endo orientation (beta-face). The X-ray crystal structure of compounds 1 and 4 show that both systems have an almost planar furanoid ring. Comparisons are made with the crystal structures of the native nucleosides (i.e., uridine and adenosine, respectively). This shows that the cyclopropane unit in 1 and the epoxide ring in 4 have virtually the same impact on the furanoid conformation, i.e., flattening of the furanoid ring is in both cases accompanied by shortening of the bonds C1'-C2' and C2'-C3' by ca. 0.03 angstrom, and expansion of the bond angles C1'-C2'-C3' and C2'-C3'-C4' by 5-6-degrees. Comparison of the crystal structures of [3.1.0]-fused nucleosides 1 and 4 with three [3.3.0]-fused nucleosides from the literature with a flattened sugar ring showed that C2'-C3' [3.1.0]-fused nucleosides display subtle structural differences, despite the fact that rotation around C2'-C3' is blocked. Secondly, a H-1 NMR conformational study on compounds 1-8 is reported. Thirdly, we have investigated whether molecular mechanics calculations (using Allinger's MM2-87 method as provided in the CHEM3D package) can be used to study the conformational properties of systems 1-8. In this respect, the structural data on 1, 4, and 8 were used to evaluate the performance of the MM2-87 method. It turns out that the molecular mechanics calculations lead to a fairly accurate picture of the structure of the modified sugar ring, while the calculated values for the torsion angles gamma and chi frequently show disparities with respect to the experimental data. It is put forward that this will be partly due to the fact that intermolecular interactions in the crystal (hydrogen bonding and base stacking) have an impact on the molecular conformation; this effect is not mimicked in our calculations.