Nicotinic acetylcholine receptors (nAChRs) are ligand gated ion channels involved in a range of important physiological processes. α-Conotoxins are short peptide toxins, which potently and specifically inhibit nAChRs. Due to their exquisite specificity for nAChR subtypes, α-conotoxins are being developed into neurochemical tools to study nAChRs and drugs to treat neurological disorders. The two main aims of this thesis are (1) to understand the interactions between α-conotoxins and nAChRs and (2) to modify the activity and specificity of α-conotoxins using computational modeling methods. These computational methods, including homology modeling, molecular dynamics (MD) simulations, and binding energy calculations, were introduced in Chapter 2.
In Chapter 3, methods to predict the binding modes and mutational energies of α-conotoxin/nAChR systems were optimized. To achieve this task, one of the most well-studied α-conotoxins, ImI, was selected as a test case. The binding mode of ImI on the α7-nAChR was modeled using homology modeling and MD simulations. This model was then employed to provide explanations to single point mutations reported in previous experimental studies. The molecular mechanics Poisson-Boltzman surface area (MM-PB/SA) and molecular mechanics generalized Born surface area (MM-GB/SA) methods were used to predict mutational energies. Strategies to predict these mutational energies were optimized for the nAChR/ImI system, and mutational energies from models minimized in explicit water and calculated using the MM-GB/SA method displayed the best correlations with the experimental values. The structure prediction method validated by mutational energy computations was further employed in Chapters 4–6.
Studies performed in Chapter 4 complement findings made in Chapter 3 on the ImI/α7-nAChR system. Several residues outside the α7-nAChR binding site significantly affect the binding affinity of ImI, but their influence could not be explained by the methods used in Chapter 3. These residues were postulated to be on the unbinding pathway of ImI, and this hypothesis was tested in Chapter 4. Non-equilibrium MD simulations, including random accelerated MD (RAMD) and steered MD (SMD) simulations, were employed to investigate the unbinding pathway of ImI. Initially, 50 RAMD simulations were performed and three unbinding pathways were identified. Only one of them was considered probable based on conformation analysis. This pathway was further divided into three subpathways, and 16 SMD simulations were carried out on each subpathway. Potential mean force calculations combined with experimental studies indicated that two subpathways were more probable to occur than the third one. Residues on these subpathways were predicted to establish specific interactions, including cation-π interactions and hydrogen bonds, to influence the unbinding kinetics of ImI.
In Chapter 5, the binding site of an analgesic α-conotoxin, Vc1.1, on α9α10-nAChR was predicted using the method developed in Chapter 3. Vc1.1 was modeled in two different binding sites: one was located between α9(+) (“+” denotes the principal subunit) and α10(-) (“-” denotes the complementary subunit), and the other was between α10(+) and α9(-). The predicted mutational energies in the two binding sites were compared to experimental data. Mutational energies calculated using the model involving the α10(+)α9(-) binding site displayed better correlations with the experimental values than models assuming an interaction in the α9(+)α10(-) binding site, suggesting that the preferred binding site of Vc1.1 mainly is α10(+)α9(-). This binding site is recommended to be used for rational modification of the activity and specificity of Vc1.1.
In Chapter 6, the specificity of three α-conotoxins for nAChR subtypes were investigated using the computational methods developed in Chapter 3. Determinants that confer specificity to α-conotoxins were investigated by predicting the structures of several systems, including ImI/α7-, α9-nAChRs, RgIA/α7-, α9α10-nAChRs, and MII/α3β2-, α6β2-nAChRs. The final results suggest that non-conserved amino acid positions could affect the specificity of α-conotoxins either by directly forming non-conserved pairwise interactions or by indirectly affecting the dynamics and conformations of the binding sites. Knowledge obtained in Chapter 6 can be used to engineer the specificity of α-conotoxins.
In Chapter 7, another aspect of α-conotoxin based drug design was investigated. In Chapter 7, one orally active cyclic α-conotoxin Vc1.1 (cVc1.1) was modified to make the oxidative folding of its disulfide bond more efficient while achieving the same global conformation. First, the contributions of the native disulfide bonds 2˗8 and 3˗16 to the stability of the peptide were evaluated using MD simulations. This analysis suggested that deletion of disulfide bond 2˗8 would introduce smaller conformational perturbation than deletion of disulfide bond 3˗16. Then different residue types were modeled at positions 2 and 8, and one analogue, cVc1.1(C2H,C8F), displayed more stable native-like conformation. The introduction of a Phe at position 8 creates a hydrophobic core and a His at position 2 participates in a network of electrostatic interactions. This analogue was chemically synthesized and its conformation was probed by NMR spectroscopy. In agreement with computational predictions, NMR studies indicated that cVc1.1(C2H,C8F) maintained the conformation of cVc1.1.