Title: 

Speaker: Dr Yoshiyui Kubota, Professor, National Institute for Physiological Sciences, Division of Multisensory Integration Systems, Okazaki, Japan

Title: Converging perisynaptic astrocytic processes from distinct astrocytes onto active dendritic segments following motor learning in the mouse primary motor cortex

Date: Monday, November 24th, 2025

Time: 11:30am - 12:30pm  

Location: THEATRE B - Tupper Building [IN-PERSON meeting]

Abstract:

Astrocytes play a key role in regulating synaptic transmission as part of the tripartite synapse. Each astrocyte typically occupies a distinct, non-overlapping domain. However, the plasticity of these domains—especially during learning-related synaptic remodeling—remains largely unknown. I will discuss our recent finding that after motor learning, perisynaptic astrocytic processes (PAPs) from multiple astrocytes converge onto the same short dendritic segment (<40 µm) of the apical tuft of a layer 5 pyramidal neuron in the mouse primary motor cortex.

Using two-photon imaging in Thy1-GFP mice trained for 8 days on a forelimb seed-reaching task, we first identified dendritic segments that exhibited high spine turnover (Sohn et al., Science Advances, 2022). Correlative light and electron microscopy (CLEM) using large-scale volume EM (vEM) data collected from these mice with automated tape-collecting ultramicrotome (ATUM) and scanning electron microscopy (SEM) revealed that these active dendritic segments were contacted by PAPs originating from 3–6 distinct astrocytes. Notably, these astrocytic processes extended directly and specifically toward each active dendritic segment. Despite the convergence of processes from multiple astrocytes at the level of dendritic segment, individual dendritic spines were typically contacted by PAPs from only a single astrocyte, indicating highly organized astrocyte-synapse interactions.

Our findings suggest the possibility that astrocytic processes dynamically reorganize in response to experience, contributing to synapse-specific modulation during motor learning.

In addition, I will briefly report our recent progress in developing EM image alignment method using image processing tool, FEABS (Finite-Element Assisted Brain Assembly System; https://github.com/YuelongWu/feabas ) and an automated dense segmentation pipeline for large-scale EM datasets.

All are welcome!