Abstract
Purpose :
Direction selective retinal ganglion cells (DSGCs) fire robustly in response to motion in one preferred direction but not in the opposite null direction. Starburst amacrine cells (SACs) mediate this computation via two mechanisms: they form asymmetric inhibitory connections onto DSGCs; and their processes display directional preference. The former is most significant during global motion, while the latter mostly contributes during local motion. We previously found that DSGCs can overcome the circuit’s anatomy and reverse their directional preference following a local repetitive visual stimulation (RVS). Here, we use this phenomenon to shed new light on the mechanisms that underlie the computation of local and global motion.
Methods :
We employed two-photon targeted patch clamp techniques from mGluR and TRHR mice, in which SACs and DSGCs express GFP, respectively. Recordings were carried while presenting the cells with various moving stimuli.
Results :
SAC processes display directional preference for local moving rings, with stronger response to centrifugal (outward from cell body) vs. centripetal (towards cell body) motion. This preference was lost following RVS. Instead, the SAC responses phase shifted, suggesting that inhibition timing, rather than amplitude, mediate DS during local motion (n=38 SACs, ***p<0.001, Student’s t-test; Figure 1A). Next, we investigated how RVS affects the computation for global motion. Prior to RVS, DSGCs display ON and OFF response phases that are tuned to the same direction. Following RVS, DSGCs displayed multiple response phases with opposed directional preferences: the early phases were tuned to the original preferred direction, while the late phases to the original null direction (n=56 DSGCs, Figure 1B).
Conclusions :
Our results reveal hidden mechanisms for the computation of direction in DSGCs that may act in orchestra to support the same computation, but can also oppose one another under certain conditions. These findings add to recent evidence demonstrating that the computations performed by anatomically-defined neuronal circuits can be altered by dynamic circuit mechanisms.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.