Summary
A bioRxiv preprint presents a synapse-resolved electron-microscopy dataset of the mouse spinal dorsal horn and a deep-learning tool for classifying sensory neuron types. Its circuit reconstruction identifies inhibitory pathways that preferentially target individual sensory subtypes while also crossing between sensory pathways.
A bioRxiv preprint describes a synapse-resolved electron-microscopy dataset of the mouse spinal cord’s dorsal horn, along with a deep-learning pipeline for identifying sensory neuron subtypes in the images. The authors used these tools to reconstruct local connections and report two patterns of inhibition: inhibitory neurons that preferentially target one sensory subtype, and connections that inhibit other sensory pathways as well.
The dorsal horn is the first major site where sensory signals entering the spinal cord are processed. Different peripheral sensory neurons detect inputs such as touch, temperature and potentially harmful stimuli. Mapping their connections can help explain how the spinal cord routes and regulates these signals before they contribute to reflexes or travel onward for perception.
A cell-type-resolved map of spinal connections
The researchers call their large-scale electron-microscopy dataset dSC1 and their classifier SPINE, short for SPinal cord Integrated NEuron classifier. Electron microscopy can reveal synaptic contacts at fine structural resolution; SPINE is designed to identify major sensory neuron subtypes directly from those images. Applying the classifier to the dataset let the team analyse the dorsal-horn wiring by sensory cell type rather than treating sensory inputs as one broad group.
The preprint reports substantial differences in synaptic architecture among physiologically distinct sensory neuron subtypes. Low-threshold mechanoreceptors—which respond to relatively gentle mechanical stimulation—formed substantially more synapses and received far more GABAergic presynaptic inhibition than the high-threshold mechanoreceptors, thermoreceptors and nociceptors examined. GABAergic presynaptic inhibition refers to inhibitory input acting at sensory-neuron axons, where it can regulate how signals are passed into spinal circuits.
Inhibition can be specific and cross-pathway
The circuit reconstruction identified five populations of spinal inhibitory neurons that form synapses onto sensory-neuron axons. Each population was reported to be preferentially driven by, and to inhibit, one sensory subtype. The authors describe this as a homotypic, modality-specific arrangement: inhibitory control is matched to a particular sensory channel.
Most of these inhibitory populations also formed connections that inhibit other sensory pathways. This second, heterotypic pattern provides a route for one sensory channel to influence the processing of another. Together, the two patterns suggest that spinal inhibition can both tune a pathway on its own and coordinate activity across modalities.
The result is a structural account of how sensory signals may be selectively regulated at an early stage of processing. The work is a mouse connectomic study, and the reported circuit motifs come from reconstructed connections; the preprint’s abstract does not describe tests of their effects on perception or behaviour. The paper was posted on bioRxiv on 1 October 2026 and is a preprint.