Abstract
Abstract: :
Purpose: To determine the role of morphology in generating direction selectivity (DS) in starburst amacrine cells. Methods: We simulated voltage responses in compartmental models generated by digitizing the morphology of starburst amacrine cells. The cell membrane was electrically passive, with constant membrane resistance Rm. Light-activated synaptic inputs were excitatory and evenly distributed across the dendrites (randomly spaced at 12 um, mean/sd = 10). DS was measured in response to both a moving bar of light and two sequential spots. We measured DS responses from the dendritic tips since that is the locus of the starburst synaptic outputs. DS was quantified as the difference in the peak response to preferred and null direction motion divided by their sum, giving a number between 0 (no DS) and 1 (max DS). Results: We found Rin increased sigmoidally with radial distance (0.5-1.5GOhm), consistent with previous reports, implying a greater response at more distal dendritic loci. Although this affected the static amplitude of signals in the dendrite it alone could not account for DS. Next we passed a bar of light across the entire cell, and found that the response to centrifugal motion (away from the soma) was greater than to centripetal motion (towards the soma). The DS-index peaked at ~.35 for stimulus velocities between 1000 and 5000 um/s. For two-spot-stimuli the DS-index was very small (~.01). To discover how independently the dendrites function, we performed the same simulations on the same cell with only one dendrite and found little DS. To investigate the role of the branching morphology we removed the side branches from the dendrites and performed the same simulations. The response to the spots once again showed a low DS-index, but for the moving bar it was higher than when the dendrite had side branches (~.07). Conclusions: DS can be generated by the morphological structure of the starburst cell. Measured at the tip of a dendrite, DS was most robust for the bar stimulus passing over the whole cell. For the preferred direction (of the dendrite), during the centripetal phase of the stimulus signals sum and traverse the soma, depolarizing the dendrite before the centrifugal stimulus reaches it. As shown by the single-dendrite model, the same centrifugal stimulus passing over the same dendrite alone does not have this advantage. In the null direction, the soma tends to shunt the signal during the centripetal phase.
Keywords: retina: proximal(bipolar, amacrine, and gangli • computational modeling • retinal connections, networks, circuitry