Summary
Perception does not depend on environment alone, but also on brain state. Work from our lab and others show
that visual responses to certain features are selectively gated depending on an animal's arousal state. What
are the circuit and synaptic bases for these state-dependent shifts in visual sensitivity? State-
dependent modulation of sensory responses has been well-described in visual cortex. Studies have also
identified behavioral modulation of responses neurons in the dorsolateral geniculate nucleus of the visual
thalamus (dLGN) of mice and primates. Remarkably, in vitro studies from the Chen lab, employing calcium
imaging and patch-clamp recordings, suggest substantial capacity for modulation of visual transmission even
earlier in the visual pathway, at the level of retinal axonal inputs to thalamus. Recently, the Andermann lab
developed methods for imaging thousands of retinal axonal boutons in thalamus of awake mice (Liang et al.,
Cell, 2018). We found that visual responses in retinothalamic boutons can be profoundly suppressed during
arousal, in a manner dependent on the boutons' visual feature preferences for stimulus location, size, motion
direction, and for luminance decreases/increases (Liang et al., Current Biology, in press). These results are
strikingly similar to the Chen lab's earlier in vitro findings of suppression of retinal ganglion cell (RGC) axonal
boutons by serotonin (5-HT). Notably, the Chen lab showed that the actions of 5-HT on RGC axons are likely
mediated by the presynaptic 5-HT1B receptor (5-HT1BR), a key receptor mediating serotonin's actions on
axon terminals throughout the brain. Preliminary data suggest that the 5-HT1BR is more strongly expressed in
axons of genetically defined RGCs with larger receptive fields. Further, our preliminary in vivo studies show
that dorsal raphe serotonergic neurons (i) are sensitive to behavioral state, (ii) send a dense and focal
projection to the dLGN, and (iii) suppress visual responses in a similar subset of RGC axons that is
suppressed by arousal. Based on these findings, the Chen and Andermann labs propose to test the hypothesis
that serotonergic inputs to the dLGN differentially suppress specific visual information in an arousal state-
dependent manner. In Aim 1, we will ask whether activity of serotonergic inputs in dLGN contribute to arousal
modulation of visual responses in RGC axons. In Aim 2, we will ask whether serotonergic inputs to dLGN
selectively gate specific channels of visual information. Finally, in Aim 3, we will ask whether serotonergic
inputs to dLGN can rapidly modify the gain and/or visual tuning of dLGN neurons. Selective suppression of
transmission at the level of retinal axons offers an efficient strategy to block non-salient retinal signals before
they are amplified by thalamocortical loops. The bridging of expertise between the Chen and Andermann labs
will establish a unified framework for understanding selective sensory processing across behavioral states at a
surprisingly early and tractable stage of visual processing. Our studies suggest that neuromodulation of retinal
axons should be considered when developing strategies for restoring vision following optic nerve damage.