Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - RETMUS (The interpretation of retinal activity by the visual thalamus.)

Teaser

Our objective is to understand visual information processing at different stages of the visual pathway. We gained a substantial knowledge about basic computation and feature extraction on the level of the retina. How visual information is introduced to higher-level visual...

Summary

Our objective is to understand visual information processing at different stages of the visual pathway. We gained a substantial knowledge about basic computation and feature extraction on the level of the retina. How visual information is introduced to higher-level visual centers, including the visual cortex, is still largely unknown. Sensory information from periphery reaches dedicated cortical areas via thalamus. Principle cells in the thalamus form the input channels to cortical areas. Originally thalamic cells were considered as relay stations transmitting information about the periphery in its original form. This model would implicate that peripheral information reaches cortical areas without being pre-processed at sub-cortical level. However recent evidence suggest that neurons in thalamus integrate input from different sensory channels and many cortical areas. We investigate this question in the context of the mouse visual system and address the following questions: is visual information from retina kept separated in different channels on the level of thalamus? Do other brain areas provide input to the principle cells of the LGN, which also receive input from the retina? What is the architecture of local inhibitory circuitry?

We revealed the finding that, even at very early stations of subcortical sensory processing, many visual signals that were long thought to be kept separated from one-another, actually get combined. Our results point to the importance of carrying out detailed mapping of cell-type-specific wiring patterns in order to understand sensory representations and processing in the visual system. We show that visual channels get combined already on the level the thalamus. We demonstrated this by carrying out a set of trans-synaptic labeling and circuit reconstruction experiments. We concluded that principle neurons in the thalamus – projecting to the visual cortex - carry out a set of different tasks. First, they serve as a relay station to cortical areas. Second, they combine information from diverse set of ganglion cells. Third, they combine information from both eyes.

We show that implications of these findings extend beyond the visual system and raise new models about brain structure, function, in particular cortical processing. The convergence of visual channels onto a single LGN cell indicates that new sensory representations is created at the level of thalamus and does not occur first and primarily within the neocortex. Neurons residing in subcortical areas like the thalamus do not only act as sensory relays by gating sensory signals traveling to the cortex but might fundamentally modify the quality of information they convey. This raises the possibility that sensory cortex receives pre-processed sensory information dynamically adjusted to the external environment we live in.

After having, in the first experimental approach, addressed how individual LGN cells combine different visual input channels from the retina, the follow-up projects will investigate how this information is integrated and modified in the LGN. This question has two main components: (1) How is visual information modified by the intra-thalamic (inhibitory) network? (2) How is visual information modified by extra-thalamic (cortical or subcortical) long-range inputs? The first question is particularly interesting for understanding basic visual processing of the LGN as a unit, which might – in the future – help to guide visual prosthetic devices to mimic the physiological patterns of visual information representation in the LGN. The second question is of outstanding importance to translate our basic knowledge of visual information processing to the situation of a complex, behaving animal. This translation requires the understanding of how visual information processing in a living animal is integrated in more complex behavioral tasks and shaped by, for example, attention and arousal.

Work performed

We investigated the building architecture of the lateral geniculate nucleus (LGN). LGN is located in the thalamus and predominantly receives input from the retina. The output of the retina is represented by the ~30 output channels, ~30 genetically identified ganglion cells that project to the LGN. Each of these ganglion cell types carries a specific feature of the visual scene. The output of the LGN is provided by the principal cells projecting to the primary visual cortex. Using monosynaptic single cell circuit tracing first we determined the heterogeneity of ganglion cell input into individual principle LGN neurons of thalamus. Second, we analyzed which other brain regions (cortical and subcortical) provide long-range input to the principle cells of LGN. Third, we analyzed the local inhibitory circuit elements synapsing onto principle cells.

An individual thalamic neuron can receive input from a homogeneous or a heterogeneous group of ganglion cells, i.e. features converge onto an individual neuron. The convergence provides the possibility for higher-level computation and feature recombination even at the level of the thalamus. It was unknown if these LGN cells only relay information from the same type of ganglion cells as suggested earlier or combine information of different types resulting in new features to emerge. We investigated this by labeling a single principal LGN cells in and their presynaptic ganglion cell partners. We developed a strategy, which ensured that we labeled a single principle LGN cell in an animal. First AAV serotype 7 was injected into the primary visual cortex. Axon terminals of principal LGN neurons were transfected with AAV, which is capable of retrograde axonal transport and therefore resulted in the fluorescent labeling and Cre-recombinase expression of principal cell bodies in the LGN. We electroporated a single fluorescently labeled cell body in each animal, up to 100 μm deep from the surface of LGN, with a fluorescent dye Alexa 568 and three plasmids: 1) one expressing the avian virus receptor in the presence of Cre (TVA); 2) one expressing the rabies glycoprotein G necessary for the transsynaptic jump of the rabies virus; 3) one expressing a fluorescent protein. We injected the rabies virus in close vicinity of the electroporated single cell.. The rabies virus could enter only the TVA-expressing single cell. Only from that single cell could the rabies virus jump to presynaptic ganglion cells. After incubation time we described the presynaptic circuit belonging to a single LGN principal neuron.

We had the following main findings:

1. Individual principal LGN cells integrate retinal input in different distinct modes
2. Relay mode cells integrate input from few retinal ganglion cells of mostly one type
3. Combination mode cells receive input from a variety of ganglion cell types
4. Binocular mode cells combine input from both eyes
5. The integration modes exhibit different degrees of cell-type specialization
6. Principle LGN cells receive input not only from the retina, but from primary visual cortex, somatosensory cortex, other thalamic nucleus and reticular nucleus

The combination of these inputs would thus establish entirely novel feature repertoire in the LGN neuron that it would in turn connect to the cortex. Our results suggest that the full extent of response properties can be more diverse than previously thought and the response properties of LGN neurons should be further investigated.

We have published our results in the following paper:

Different Modes of Visual Integration in the Lateral Geniculate Nucleus Revealed by Single-Cell-Initiated Transsynaptic Tracing. Rompani SB, Müllner FE, Wanner A, Zhang C, Roth CN, Yonehara K, Roska B. Neuron. 2017 Mar 22;93(6):1519.

Final results

One of our long-term research objectives is to provide ideas how to restore visual functions in blind patients in the future. To achieve that we need to understand the computational architecture at three main stages of the visual pathway: retina, thalamus and visual cortex. Our aim is to build upon the knowledge obtained in this program about the visual part of the thalamus to further develop vision restoration therapies for the visual pathway.
Retinal degeneration is one of the leading causes of blindness in the Western World. In several forms of retinal degeneration ganglion cells are affected, like in glaucoma and optic neuropathies. These disease conditions affect the retina output channels preventing visual information reaching higher visual brain centers while the rest of the visual pathway remains intact. Therefore we can aim to restore some useful visual perception by reintroducing visual information directly into higher visual centers, the thalamus for instance. For that purpose we target specific cell types with optogenetic tools that allows us to reactivate the visual pathway. Gaining knowledge about precise thalamic circuitry would help us introducing visual information directly and effectively into the sensory cortex via activation of principle thalamic neurons. For Sensory restoration, which is of major relevance for regaining autonomy, will improve the quality of daily life of patients in the long term.