Research

Direction selective retinal ganglion cells respond strongly to an image moving in one direction (the “preferred” direction) and weakly to an image moving in the opposite, “null”, direction. The asymmetric computation is thought to be hardwired, arising from asymmetric inhibitory inputs from starburst amacrine cells (SACs). Yet, we have discovered that the directional preference of a neuron can be altered and reversed by 180 degrees following a short repetitive visual stimulation. This reversal is robust and long lasting, and occurs not only in juvenile but also in adult mice.

Direction selective ganglion cells (DSGCs) reverse their directional preference

Directional tuning of a DSGC before (left) and after (right) repetitive stimulation with 40 sec of drifting grating (shown in center), revealing reversal of directional preference. Polar plot represents number of spikes in response to 3 sec gratings drifting in 12 directions. Red arrow indicates the preferred direction. Traces show examples of 0.5 sec activity.

Adopted from Rivlin-Etzion et al., Neuron 2012

To understand how DSGCs overcome the circuit’s anatomy and reverse their directional preference, we record from neurons in the direction selective circuit using different patch-clamp techniques and in response to various moving stimuli. A major focus is the SAC that is known to mediate the directional response in DSGC. Our data sheds new light on the multiple mechanisms that underlie direction selectivity and how they can act in orchestra or oppose one another.

The visually-induced functional switches are not restricted to direction selective ganglion cells. The ON and OFF retinal pathways that signal light increments and decrements, respectively, are well known for their segregate parallel processing. Yet, we and others find that subtypes of retinal cells can switch their polarity following repetitive stimulation or following changes in ambient light levels.

An RGC’s polarity preference changes with light illumination

Peri-stimulus time histograms (PSTHs) of transient-Off-alpha RGC in response to a dark spot at low (left) and high (right) light intensities (64 and 64x104 R*/rod/sec, respectively; same contrast), revealing both On and Off responses in low illumination but only an Off response in high illumination. Dark bars indicate spot period.

Right: Z-stack projections (top) and side view (bottom) reveal morphology of the RGC.

We use two-photon calcium imaging and electrophysiology to determine the prevalence of retinal dynamic computing and identify types of RGCs that tend change their function. In addition, we explore the various conditions and dimensions over which retinal computations are dynamic.

Stimulus characteristics of the mouse’s visual field differ above and below the skyline. Do RGCs change their functional properties with their retinal location to allow better representation of the different stimulus characteristics?

Transient-Off-alpha-RGCs gradually change their response properties along the dorsal-ventral axis

Left: (Top) Positions of 22 transient-Off-αRGCs recorded from across the retina. Response durations to a 400 µm spot are colored coded. Cardinal axes are marked in the center. D: dorsal, V: ventral, T: temporal, N: nasal.

(Bottom) Plot of response duration as a function of position along the ventral-dorsal axis.

Right: PSTHs of 3 representative transient-Off-αRGCs located in dorsal (top), center (middle) and ventral (bottom) retina.

Adopted from Warwick et al., Current Biology 2018

Our findings challenge the current belief that RGCs of the same type exhibit the same light responses regardless of retinal location, and suggest that networks underlying RGC responses may change with location to enable optimized sampling of the visual image. We use two-photon targeted voltage-clamp recordings, pharmacology, and genetic manipulations to uncover the neuronal networks that mediate this gradual change.

The retina is known for its adaptive response to light and contrast levels. These adaptations lead to changes in sensitivity but not to changes in function. The finding that retinal neurons can change their core computations was unexpected and is a main focus of our lab’s research.

We pioneer new techniques and approaches to decipher the transfer function between the retina and the visual thalamus (LGN), and to determine how retinal dynamics are decoded by their targets.