Thalamic activation of the visual cortex at the single-synapse level

How does the mammalian primary visual cortex (V1) achieve orientation selectivity? Specifically, the authors investigated whether orientation-tuned neurons in layer 4 (L4) receive inputs from orientation-tuned thalamic neurons or from non-selective, canonical thalamic neurons, as originally proposed by the Hubel and Wiesel (H&W) feedforward model.

To resolve this, the researchers employed a sophisticated suite of techniques in vivo. They used two-photon glutamate imaging and calcium imaging to monitor activity at individual synapses on L4 pyramidal neurons. By combining these imaging methods with optogenetic cortical silencing, they were able to distinguish between thalamocortical (TC) inputs and corticocortical (CC) inputs. Furthermore, they used single-cell–initiated rabies virus tracing to map the presynaptic connectivity of these neurons and analyzed the spatial arrangement of thalamic receptive fields using drifting-bar visual stimulation.

The study revealed that TC synapses onto orientation-tuned L4 neurons are non-selective and originate from canonical, non-orientation-tuned thalamic neurons. In contrast, CC inputs were found to be highly orientation-selective, often matching the preferred orientation of the postsynaptic neuron. Crucially, the researchers observed that the receptive fields of these TC inputs are spatially arranged in a way that aligns with the preferred orientation of the L4 neuron, providing direct experimental evidence for the H&W feedforward model. Additionally, the authors noted that TC synapses lack the postsynaptic calcium signals typically seen at CC synapses, suggesting distinct synaptic properties and NMDA receptor expression profiles at these two types of inputs.

This work provides a definitive synaptic-level resolution to a long-standing debate in visual neuroscience. By mapping the functional wiring of individual synapses, the authors confirm that the emergence of orientation selectivity in the cortex is primarily driven by the precise spatial convergence of non-selective thalamic inputs, rather than by the direct input of already-tuned thalamic neurons. This advances our understanding of how sensory information is processed and transformed within the cortical circuit.