Mouse dLGN receives input from a diverse population of retinal ganglion cells with limited convergence

In the mammalian brain RGCs transfer information to more than 50 different regions, including the dorsal lateral geniculate nucleus (dLGN) in the thalamus. The dLGN extensively connects with the primary visual cortex, and is part of the pathway that gives rise to conscious visual perception or sight.

How then are visual inputs from the retina processed in the thalamus, the first stop in the brain’s visual pathway? To answer this question, studies have looked at “RGC convergence” – the number and diversity of RGCs that project to individual dLGN thalamocortical (TC) neurons. Low convergence would suggest a more relay-like transmission, whereas high convergence implies that a more complex transformation or integration of visual information may take place in the thalamus, before reaching the cortex.

Prior physiological work in the dLGN across mammalian species suggested only limited RGC convergence onto target dLGN cells. However, more recent structural electron microscopy imaging, transsynaptic tracing, and functional calcium imaging of retinal axonal boutons in the mouse dLGN all showed that much higher RGC convergence is possible, with up to 90 RGCs from multiple classes innervating a single dLGN neuron. In their recent preprint, Román Rosón et al. also address these questions.

To look into how this wealth of information is processed by the thalamus, the authors were interested in the degree of functional RGC convergence onto single TC neurons in the dLGN. Here, functional convergence is defined as the number of different RGC types that meaningfully drive the responses of a TC cell to visual stimuli.

To this end, they first performed extracellular single-unit recordings of geniculate neurons in head-fixed mice, which were presented the same set of stimuli as the retinas. From this data, they observed that both TC and RGC responses showed similar diversity.

Finally, the study used a linear model to predict dLGN responses as a weighted sum of excitatory RGC inputs. This modelling implies that on average just 2 different RGC groups drive the response of a dLGN TC neuron. The estimate, however, varied depending on what threshold was chosen for inputs to be considered functionally relevant. Taken together, this would therefore suggest a limited functional retinal convergence in the visual thalamus.

Interestingly, the model also showed that RGC classes overrepresented in the dLGN projecting RGC population contributed significantly to dLGN TC neuron responses. This would suggest that the information these RGCs relay is particularly significant for subsequent visual computations.