Crystal structure of a new benzoic acid inhibitor of influenza neuraminidase bound with a new tilt induced by overpacking subsite C6

BACKGROUND: Influenza neuraminidase (NA) is an important target for antiviral inhibitors since its active site is highly conserved such that inhibitors can be cross-reactive against multiple types and subtypes of influenza. Here, we discuss the crystal structure of neuraminidase subtype N9 complexed with a new benzoic acid based inhibitor (2) that was designed to add contacts by overpacking one side of the active site pocket. Inhibitor 2 uses benzoic acid to mimic the pyranose ring, a bis-(hydroxymethyl)-substituted 2-pyrrolidinone ring in place of the N-acetyl group of the sialic acid, and a branched aliphatic structure to fill the sialic acid C6 subsite. RESULTS: Inhibitor 2 {4-[2,2-bis(hydroxymethyl)-5-oxo-pyrrolidin-1-yl]-3-[(dipropylamino)methyl)]benzoic acid} was soaked into crystals of neuraminidase of A/tern/Australia/G70c/75 (N9), and the structure refined with 1.55 Å X-ray data. The benzene ring of the inhibitor tilted 8.9° compared to the previous compound (1), and the number of contacts, including hydrogen bonds, increased. However, the IC(50) for compound 2 remained in the low micromolar range, likely because one propyl group was disordered. In this high-resolution structure of NA isolated from virus grown in chicken eggs, we found electron density for additional sugar units on the N-linked glycans compared to previous neuraminidase structures. In particular, seven mannoses and two N-acetylglucosamines are visible in the glycan attached to Asn200. This long, branched high-mannose glycan makes significant contacts with the neighboring subunit. CONCLUSIONS: We designed inhibitor 2 with an extended substituent at C4-corresponding to C6 of sialic acid-to increase the contact surface in the C6-subsite and to force the benzene ring to tilt to maximize these interactions while retaining the interactions of the carboxylate and the pyrolidinone substituents. The crystal structure at 1.55 Å showed that we partially succeeded in that the ring in 2 is tilted relative to 1 and the number of contacts increased, but one hydrophobic branch makes no contacts, perhaps explaining why the IC(50) did not decrease. Future design efforts will include branches of unequal length so that both branches may be accommodated in the C6-subsite without conformational disorder. The high-mannose glycan attached to Asn200 makes several inter-subunit contacts and appears to stabilize the tetramer.