Astrocytes refine cortical connectivity at dendritic spines

Abstract

During cortical synaptic development, thalamic axons must establish synaptic connections despite the presence of the more abundant intracortical projections. How thalamocortical synapses are formed and maintained in this competitive environment is unknown. Here, we show that astrocyte-secreted protein hevin is required for normal thalamocortical synaptic connectivity in the mouse cortex. Absence of hevin results in a profound, long-lasting reduction in thalamocortical synapses accompanied by a transient increase in intracortical excitatory connections. Three-dimensional reconstructions of cortical neurons from serial section electron microscopy (ssEM) revealed that, during early postnatal development, dendritic spines often receive multiple excitatory inputs. Immuno-EM and confocal analyses revealed that majority of the spines with multiple excitatory contacts (SMECs) receive simultaneous thalamic and cortical inputs. Proportion of SMECs diminishes as the brain develops, but SMECs remain abundant in Hevin-null mice. These findings reveal that, through secretion of hevin, astrocytes control an important developmental synaptic refinement process at dendritic spines.

DOI: http://dx.doi.org/10.7554/eLife.04047.001

eLife digest

The central nervous system—which is made up of the brain and spinal cord—processes information from all over the body. The information travels through cells called neurons, which connect to each other at junctions called synapses. A single neuron can receive information from many different places because it is covered with protrusions known as dendritic spines that enable it to form synapses with a variety of other neurons.

In recent years, it has become apparent that brain cells other than neurons can influence synapse formation. The most abundant cells in the central nervous system are star-shaped cells known as astrocytes, which secrete molecules that control the timing and extent of synapse formation. Many previous studies on synapses have used a type of neuron found in the eye—called retinal ganglion cells—because these cells can be purified and grown in the laboratory in the absence of astrocytes. Under these conditions, they form very few synapses. However, in the presence of astrocytes the retinal ganglion cells form many more synapses, which is thought to be due to a protein called hevin and several other proteins that are secreted by the astrocytes.

Risher et al. studied a region of the brain called the cerebral cortex in mice that were missing hevin. In the cortex of normal mice, the neurons generally form synapses with other neurons within the cortex, or with neurons from other parts of the brain that send long-distance projections into the cortex. The experiments revealed that fewer of these long-distance synapses formed in the cortex of the mice missing hevin compared to normal mice. When hevin was injected directly into the brains of the mice, more long-distance synapses were formed.

Using a technique called three-dimensional electron microscopy, Risher et al. examined the structure of the synapses. In mice missing hevin, the synapses were much smaller and the dendritic spines were thin and long, indicating that they were not fully grown. The images also show that in normal mice, the dendritic spines often have multiple synapses when the animal is young, but many are lost as the brain matures so that only a single synapse remains in each dendritic spine. However, multiple synapses persist in the dendritic spines of mice lacking hevin, which could lead to competition between short and long distance synapses and may contribute to neurological diseases.

These results indicate that astrocytes are crucial for controlling the formation of synapses in dendritic spines. In humans, defects in hevin have been implicated in autism, schizophrenia and other neurological conditions. Future studies will seek to determine the precise role of astrocytes in these conditions, which may help us to develop new therapies.

DOI: http://dx.doi.org/10.7554/eLife.04047.002