Predictions of secondary structure and solvent accessibility of the light chain of the clostridial neurotoxins.

Predictions were made of the secondary, two-dimensional (2-D) structures and side-chain solvent accessibilities of the light (L) chains of the clostridial neurotoxins (botulinum neurotoxin serotypes A-G and tetanus neurotoxin). An artificial neural network was used to make these predictions from a multiple alignment of their primary structures and was the approach used in making successful predictions for the C-fragments of these neurotoxins (Lebeda et al., J. Prot. Chem., 17:311, 1998). We also exploited the fact that the L-chains are Zn-dependent proteases. Although no other metalloproteases were found to be sequentially homologous to these neurotoxin L-chains, a sequence clustering algorithm showed that several bacterially derived Zn-dependent proteases, including thermolysin, were the most similar. A 2-D structure topology map for the type A L-chain was constructed by using thermolysin as a design template. As in thermolysin, the region containing the Zn-binding sequence motif, which is part of the active site in these neurotoxins, was predicted to be minimally solvent accessible. On the other hand, the locations of residues with highly exposed side chains were predicted to occur in non-periodic structure elements. Together, these 2-D structure and solvent accessibility predictions can be used to identify important solvent-exposed regions of the L-chain. These regions may include sites that interact with residues of the neurotoxin heavy chain, sites that bind to vesicle-docking substrates or sites that form antibody epitopes.