Abstract
The passive membrane properties of the tangential cells in the fly lobula plate (CH, HS, and VS cells, Fig. 1) were determined by combining compartmental modeling and current injection experiments. As a prerequisite, we built a digital base of the cells by 3D-reconstructing individual tangential cells from cobalt-stained material including both CH cells (VCH and DCH cells), all three HS cells (HSN, HSE, and HSS cells) and most members of the VS cell family (Figs. 2, 3). In a first series of experiments, hyperpolarizing and depolarizing currents were injected to determine steady-state I-V curves (Fig. 4). At potentials more negative than resting, a linear relationship holds, whereas at potentials more positive than resting, an outward rectification is observed. Therefore, in all subsequent experiments, when a sinusoidal current of variable frequency was injected, a negative DC current was superimposed to keep the neurons in a hyperpolarized state. The resulting amplitude and phase spectra revealed an average steady-state input resistance of 4 to 5 M Omega and a cut-off frequency between 40 and 80 Hz (Fig. 5). To determine the passive membrane parameters R(m) (specific membrane resistance), R(i) (specific internal resistivity), and C-m (specific membrane capacitance), the experiments were repeated in computer simulations on compartmental models of the cells (Fig. 6). Good fits between experimental and simulation data were obtained for the following values: R(m)=2.5 k Omega cm(2), R(i)=60 Omega cm, and C-m=1.5 mu F/cm(2) for CH cells; R(m)=2.0 k Omega cm(2) R(i)=40 Omega cm, and C-m=0.9 mu F/cm(2) for HS cells; R(m)=2.0 k Omega cm(2), R(i)=40 Omega cm, and C-m=0.8 mu F/cm(2) for VS cells. An error analysis of the fitting procedure revealed an area of confidence in the R(m)-R(i) plane within which the R(m)-R(i) value pairs are still compatible with the experimental data given the statistical fluctuations inherent in the experiments (Figs. 7, 8). We also investigated whether there exist characteristic differences between different members of the same cell class and how much the exact placement of the electrode (within+/-100 mu m along the axon) influences the result of the simulation (Fig. 9). The membrane parameters were further examined by injection of a hyperpolarizing current pulse (Fig. 10). The resulting compartmental models (Fig. 11) based on the passive membrane parameters determined in this way form the basis of forthcoming studies on dendritic integration and signal propagation in the fly tangential cells (Haag et al., 1997; Haag and Borst, 1997).
