Cross-reactive antibody S139/1 docks to hemagglutinin H3 subtype using RDOCK protein-protein docking suite, in research carried out by Alexandra Delaney at Universiti Sains Malaysia.


Victor Chu, Pek Ieong, Alexandra Delaney and Howard Li, PRIME 2011 (China, Malaysia and Taiwan)

The influenza virus is still a widely prevalent threat to global public health and the world-wide economy. During the summer of 2011, four UCSD PRIME students worked on various aspects of influenza biology at three host sites: Victor Chu at CNIC in Beijing; Pek Ieong and Alexandra Delaney at USM in Penang; and Howard Li at National Taiwan University in Taipei.

Chu continued virtual screening studies using new target sites in hemagglutinin (HA), while Ieong and Delaney explored the glycobiology and immunology of HA. Meanwhile, Li focused on novel scaffolds for inhibitor design against neuraminidase (NA).

Li describes the novel finding that adamantane, a known influenza inhibitor rendered obsolete due to viral resistance, may contain a scaffold effective for synthesis of new inhibitors against drug resistant forms of NA.NA is an enzyme that mediates the budding and release of daughter viral particles from host cells. It is essential for the spread of influenza infections, and thus a key target for antiviral drug development. At NTU, Li investigated a scaffold-based, fragment-growing method to identify molecular scaffolds for Group 1 neuraminidase inhibition. Premising this study are two assumptions: 1) effective inhibitors possess not only an optimal combination of functional groups but also a fundamental molecular geometry, dictated by a core scaffold, that allows the molecule to better access the binding site; and 2) the potential of a molecular scaffold to be developed into an effective inhibitor can be explored through fragment-growing algorithms.

An initial set of basic geometric scaffolds, along with the core ring structures of Zanamivir and other known NA inhibitors, were inputted into AutoGrow (the fragment-growing application used in his study). AutoGrow randomly mutates an initial input molecule with substituents from a fragment library. Using AutoDock Vina as a scoring function by docking mutated compounds to a known oseltamivir-resistant NA protein structure, AutoGrow mutations that lead to a greater binding affinity are preserved for the next generation and cycle of mutation; selection continues until an effective inhibitor eventually evolves. After eight cycles, all scaffolds showed a general increase in binding affinity; however, some clearly showed more successful evolutionary trajectories and evolved into ligands of significantly higher binding affinity. Adamantane was thus identified by this study as a novel scaffold for possible development into a potential inhibitor. Analyzing the substituents of adamantane derivatives revealed a consistent trend: the top endpoint molecules of multiple parallel evolutions all resulted in the attachment of a sulfate-containing branch, as well as an alcoholic branch exactly four bonds away. As validation for this method, ligands derived from the core scaffold of Zanamivir had nearly identical fragment attachment sites as the original Zanamivir ligand.

At USM, Alexandra Delaney chose a challenging topic based on her interest in immunology. She wanted to study antigen recognition, an integral aspect of the humoral immune response and the potential benefits of antibody engineering. Despite the large body of biological investigations regarding antigen-antibody interactions, there is a noticeable absence of in silico protein docking analysis to complement these studies. She developed in silico protocols to construct a model for a cross-reactive antibody, S139/1, and predict the free energy of binding (FEB) of S139/1 to HA. The computational workflow allows one to design high-affinity, neutralizing antibodies without the use of biological expression libraries. She used techniques in homology modeling, protein-protein docking and refinement, and molecular dynamics (MD) to construct, dock, and predict the binding energy of S139/1 to hemagglutinin H3 subtype. Initially, MODELLER v9.2, in conjunction with WAM web server and Rosetta Antibody structure prediction server, were used to generate homology models of S139/1; S139/1 was then docked to H3 using the Accelrys Discovery Studio (DS) v2.5 protein-protein docking suite (ZDOCK, RDOCK, and ZRANK). Finally, MD simulations were completed to obtain the FEB more accurately. The ZDOCK and RDOCK results indicated that it is possible to predict the FEB of protein-protein complexes relative to negative and positive controls, and engineer theoretical antibodies that bind to HA. She is planning to follow up her summer research using the supercomputing facilities at UCSD.

Ultimately, Delaney would like to identify candidate high-affinity, neutralizing antibodies that can be synthesized for further biological investigations of their inhibition of influenza infection. These antibody-engineering techniques would complement existing small molecule-based drug therapies, as well as prophylactic vaccines.

PARTICIPATING RESEARCHERS: CNIC: Kai Nan, Kevin Dong; USM: Habibah Wahab, Sybing Choi, Mohammed Yusuf; NTU:Jung-Hsin Lin; SDSC/UCSD: Wilfred Li.