Synthesis of tamiflu and its phosphonate congeners possessing potent anti-influenza activity.

Influenza remains a major health problem for humans and animals.1 At present, four drugs are approved for influenza prophylaxis and treatment:2 amantadine and rimantadine act as the M2 ion channel blockers, whereas Tamiflu (the phosphate salt of oseltamivir ethyl ester) and Relenza (zanamivir) inhibit the activity of neuraminidase (NA). The NA inhibitors (NAIs) are designed to have (oxa)cyclohexene scaffolds to mimic the oxonium transitionstate in the enzymatic cleavage of sialic acid.3 Tamiflu (1, shown in Scheme 1) is an orally administrated anti-influenza drug.4 On hydrolysis by hepatic esterases, the active carboxylate, oseltamivir (2, also known as GS4071), is exposed to interact with three arginine residues (Arg118, Arg292, and Arg371) in the active site of NA.3 The phosphonate group is generally used as a bioisostere of carboxylate in drug design.5 In comparison with the carboxylateguanidinium ion pair, a phosphonate ion exhibits stronger electrostatic interactions with the guanidinium ion. Our preliminary molecular docking experiments (Figure 1) using the known N1 crystal structure (PDB code: 2HU4)3c reveal that the putative phosphonate inhibitor 3a indeed binds strongly with the triarginine residues of NA, in addition to other interactions exerted by the C3-pentyloxy, C4-acetamido, and C5-amino groups in the binding pocket similar to the NA-oseltamivir complex. Because the previously reported methods4 for the synthesis of oseltamivir/ Tamiflu are not amenable to exchange of the C-1 carboxyl group to a phosphonate group, we thus explored a novel approach to the synthesis of both oseltamivir/Tamiflu and the phosphonate congeners using D-xylose as an appropriate chiral precursor (Scheme 1). In brief, our present synthetic method is straightforward to culminate in an enantioselective synthesis of Tamiflu, oseltamivir, the phosphonate congener, and the guanidine analogues with reasonably high yields (5.2-13.5%). An intramolecular HornerWadsworth-Emmons reaction was carried out to furnish the cyclohexene carboxylate 8a and phosphonate 8b. On treatment with diphenylphosphoryl azide according to Mitsunobu’s method,6 the hydroxyl group in 8a/8b was successfully substituted by an azido group with the inversed configuration. The hazardous reagent of sodium azide was avoided in this procedure. This synthetic scheme allows late functionalization, which makes it attractive from a medicinal chemistry point of view. The greater potencies of the phosphonate congeners, 3 (namely Tamiphosphor) versus oseltamivir 2 and guanidine 13b versus 13a, were observed in the wild-type neuraminidases of H1N1 and H5N1 influenza viruses (Table 1). Both compounds 3 and 2 are significantly less potent toward the NAI resistant mutants of H274Y7 than the wild-type enzymes. Nevertheless, the phosphonate compound 13b is an effective inhibitor that inhibits both mutant enzymes at low nM concentrations. Compounds 14a and 14b, which lack the † Academia Sinica. ‡ National Taiwan University. § The Scripps Research Institute. Figure 1. Molecular models of oseltamivir 2 (left panel) and the phosphonate compound 3a (right panel) in the active site of influenza virus neuraminidase (N1 subtype). The complex of the phosphonate compound 3a has more extensive hydrogen bonding interactions (8 pairs ligand-NA H-bonds) with key residues in the NA active site than the oseltamivir-NA complex (6 pairs ligand-NA H-bonds)