Differential temporal filtering in the fly optic lobe

Abstract

Visual interneurons come in many different flavors, representing luminance changes at one location as ON or OFF signals with different dynamics, ranging from purely sustained to sharply transient responses. While the functional relevance of this representation for subsequent computations like direction-selective motion detection is well understood, the mechanisms by which these differences in dynamics arise are unclear. Here, I study this question in the fly optic lobe. Taking advantage of the known connectome I simulate a network of five adjacent optical columns each comprising 65 different cell types. Each neuron is modeled as an electrically compact single compartment, conductance-based element that receives input from other neurons within its column and from its neighboring columns according to the intra- and inter-columnar connectivity matrix. The sign of the input is determined according to the known transmitter type of the presynaptic neuron and the receptor on the postsynaptic side. In addition, some of the neurons are given voltage-dependent conductances known from the fly transcriptome. As free parameters, each neuron has an input and an output gain, applied to all its input and output synapses, respectively. The parameters are adjusted such that the spatio-temporal receptive field properties of 13 out of the 65 simulated neurons match the experimentally determined ones as closely as possible. Despite the fact that all neurons have identical leak conductance and membrane capacitance, this procedure leads to a surprisingly good fit to the data, where specific neurons respond transiently while others respond in a sustained way to luminance changes. This fit critically depends on the presence of an H-current in some of the first-order interneurons, i.e., lamina cells L1 and L2: turning off the H-current eliminates the transient response nature of many neurons leaving only sustained responses in all of the examined interneurons. I conclude that the diverse dynamic response behavior of the columnar neurons in the fly optic lobe starts in the lamina and is created by their different intrinsic membrane properties. I predict that eliminating the hyperpolarization-activated current by RNAi should strongly affect the dynamics of many medulla neurons and, consequently, also higher-order functions depending on them like direction-selectivity in T4 and T5 neurons. Author summaryVisual interneurons come in many different flavors, representing luminance changes at one location as ON or OFF signals and with different dynamics, ranging from purely sustained to sharply transient. While the functional relevance of this representation for subsequent computations, like direction-selective motion detection, is well understood, the mechanism by which these differences in dynamics arise is unclear. Here, I study this question by using the connectome of the fly optic lobe and simulating the network of interneurons in a biophysically plausible way. Adjusting the input and the output gain of each neuron such that a subset of neurons (those where experimental data exist) matches the response kinetics of their real counterparts, I identify a particular voltage-gated ion channel present in some of the first-order interneurons as being critical for the transient response behavior of postsynaptic neurons. This study, therefore, predicts that eliminating this current from the circuit should strongly affect the response kinetics in downstream circuit elements and destruct the computation of direction selectivity.