Application of the log-normal model for long term high affinity antibody/antigen interactions using Bio-Layer Interferometry

For more than 50 years, optical biosensors have been used to measure bio-molecular interactions. The most frequently applied binding model to fit biosensor data is the simple 1:1 binding model which requires the stabilization of the association phase to the equilibrium Req and the stabilization of the dissociation phase to the equilibrium zero. However, due to technical limitations many published biosensor measurements are finished before these requirements are fulfilled. In the present study, a long term binding interaction analysis with a monoclonal antibody, namely IgG 2F5 and UG37 a specific antigen with a promising biosensor platform, the Bio-Layer Interferometry, was performed. Data fitting with the simple 1:1 binding model to the association phase was inappropriate and the fitted parameters varied with the concentration and time, which contradicts the theory of the simple 1:1 binding model. Furthermore, extrapolation of the fits with individual times spans compared to 100 % of the obtained data systematically underestimated the actual observed binding curve. Interestingly, an alternative model based on the cumulative distribution function of the log-normal probability distribution remedied the aforementioned problems allowing $hbox {K}_mathrm{L}$ (which is the analog to the affinity constant $hbox {K}_mathrm{D}$ ) to be estimated. We further demonstrate that this model fits the biosensor data far better and is essentially less affected by the stabilization of the association phase to the equilibrium (Req) and the stabilization of the dissociation phase to the equilibrium zero. Finally, extrapolation with the log-normal model predicts the actually observed binding curve in a proper manner.