Fig. 3

In silico FCS simulations. (A) Binary image shows a slice of the OT region marked with examples of a void and channel detection region for FCS experiments by red circles. Detection region is a sphere of radius 0.1 ?m. On the right is shown an example output of the "intensity" profile by counting the number of particles in the detection region at each simulation time step, which is 0.04 ms. The black dashed line is the mean number of particles across the simulation related to in the autocorrelation plots. (B) FCS experiments simulated for different regions (void and channels) of OT images for the cases without binding (left) and receptor bindings on cell surfaces (right). Mean temporal autocorrelation (solid and dashed lines) with standard deviation (shaded regions) of the "intensity" profiles for increasing lag times is plotted on a log scale and fitted with 3D diffusion model (Eq. (9)). Left figure shows autocorrelation curves of different colors for diffusion coefficients 10?80 ?m2/s for voids (solid lines) and channels (dashed lines) without binding events. Linear regression (inset) shows that measured approximately recovers the input diffusion coefficient in this parameter regime, with channel region showing slightly more slowdown due to hindered diffusion. The right figure shows mean autocorrelation curves for voids and channels with receptor bindings at cell surfaces. This shows two populations, slow- and fast-diffusing molecules in the system, localized in the channel and void regions, respectively. For FCS in voids (purple), the measured is m2/s, whereas FCS data in channels were fitted with a two-species model, in which of particles were a slow-diffusing component with fitted m2/s. (C) FCS experiment is simulated in voids for different architectures of the zebrafish brain for ?m2/s, and curves for ?m2/s for simulation on different architectures are shown here. measurements are of similar magnitude as the input diffusion coefficient , showing that FCS measurements recover the input diffusion coefficient of the simulations. This shows that surrounding tortuosity does not affect results when measuring local diffusion coefficient.