Cells called astrocytes promote and maintain neuronal function. The discovery that astrocytes vary in their gene expression, protein levels, cellular structure and function suggests that they are specialized to support distinct circuits.
Approximately one-third of cells in the mammalian central nervous system (CNS) are star-shaped astrocytes. These non-neuronal cells have many vital roles in CNS function1. For example, they instruct the formation, maturation and elimination of synaptic connections between neurons during development. They are involved in maintaining the blood–brain barrier and in regulating blood flow. They mediate neuronal uptake and the recycling of neurotransmitter molecules, and they provide nutritional and metabolic support to neurons. Are these functions uniformly executed by astrocytes throughout the CNS, or are there regional differences? Writing in Neuron, Chai et al.2 present the first comprehensive study to demonstrate brain-region-specific astrocyte specializations.
Although there has long been an understanding of the variability between different neuronal populations, our understanding of astrocyte heterogeneity is still in its infancy. It was originally assumed that all astrocytes are the same, but several studies1,3,4 performed during the past decade have reported differences in both cell shape and the enrichment of proteins in astrocytes in different brain regions. Whether these differences translate into functional specializations has not been clear.
Supporting the idea of functional heterogeneity in the brain, astrocytes purified from different areas of the spinal cord are heterogeneous, differentially expressing guidance cues that direct neurons to form distinct sensorimotor circuits5. Moreover, gene-expression differences have also been reported in the brain — for example, the typical astrocyte marker gene GFAP is variably expressed in different cortical regions1,3,4,6.
Working with mice, Chai and colleagues analysed two mature brain regions that have functionally distinct neuronal circuits — the hippocampus and the striatum. The hippocampus is necessary for establishing short-term, long-term and spatial memory, whereas the striatum integrates excitatory and inhibitory signals from many parts of the brain to coordinate voluntary movement. Are astrocytes in these regions specialized to meet the functional requirements of these distinct neuronal circuits?
The authors first used a combination of imaging techniques to analyse the morphology of astrocytes (Fig. 1). They found that astrocyte density was similar in the two regions, but that striatal astrocytes extended processes to cover larger territories. By contrast, hippocampal astrocytes make contact with nearly twice as many synapses as do striatal astrocytes. Hippocampal astrocytes therefore probably have the capacity to integrate more neuronal signals.
Do these differences enable astrocytes to modulate neuronal activity in a circuit-specific manner? Chai et al. addressed this question by investigating whether astrocytes in each brain region are functionally different. Unlike neurons, which signal through electrical impulses, astrocyte communication involves the transfer of calcium ions (Ca2+) through junctions between adjacent cells. The authors used genetically encoded calcium indicators (proteins that fluoresce in response to an increase in Ca2+ levels) to determine the calcium-signalling dynamics in various regions of individual astrocytes. This revealed that Ca2+ signalling differed between hippocampal and striatal astrocytes.
Although widely studied, the function of astrocytic Ca2+ signalling is still under debate. One hypothesis is that increased neuronal activity causes elevations in astrocyte Ca2+ levels, which, in turn, triggers release of the neurotransmitter glutamate from astrocytes7. Chai and colleagues examined this possibility in astrocytes in both brain regions. They found little evidence for expression of the molecular machinery required to release glutamate, and using a genetically encoded glutamate sensor they could not detect Ca2+-dependent glutamate release from astrocytes in brain slices. Together, these data argue against the idea of glutamate release from astrocytes in vivo7,8.
Are regional-specific differences in astrocytes genetically encoded, or are they the result of differences in the specific brain region in which they developed? To address this question, the authors used RNA sequencing to generate a transcriptome database — a profile of all the messenger RNA molecules present in cells — for adult astrocytes from each brain region. To do this, they used a genetically engineered mouse in which a subunit of the ribosomal machinery (which translates mRNA into protein) that had been modified with a molecular tag was expressed specifically in astrocytes9,10. The tag allows astrocyte-specific extraction of mRNA. Analysis revealed differential expression of 2,818 genes between brain regions, some of which the authors validated separately using a range of approaches, including analysis of protein abundance.
Strikingly, Chai and colleagues report that the gene that encodes ALDH5A1 — a protein that degrades the inhibitory neurotransmitter GABA — was enriched in striatal astrocytes. Perhaps ALDH5A1 enables these astrocytes to efficiently control GABA levels in the striatum, which is densely packed with GABA-releasing neurons. Furthermore, the authors found that the gene that encodes the protein μ-crystallin is highly enriched in striatal astrocytes. This is intriguing because μ-crystallin levels decrease in people with Huntington's disease and in mouse models of this disorder11. Thus, the authors' finding raises the question of whether astrocytes could have a pivotal role in the pathogenesis of this and other neurodegenerative diseases.
These findings will undoubtedly influence future research into astrocyte biology. Not only do they convincingly demonstrate that astrocytes are regionally specialized and probably have circuit-specific functions, they also raise many questions. For instance, how does astrocyte specialization aid information processing in the brain? One could speculate that the larger number of synapses contacted by astrocytes in the hippocampus allows for integration of the complex array of neuronal signals required for memory formation and retrieval. How is astrocyte diversity achieved? Perhaps the astrocytes in each region are produced from different progenitor pools, as in the spinal cord3,5. Maybe these differences are intrinsically specified, or driven by the local environment. It will also be instructive to determine whether subregional astrocyte heterogeneity exists, allowing the cells to discriminate between neuronal inputs within an anatomical circuit to further optimize neuronal function.
Ultimately, Chai and colleagues' combined experimental approach, coupled with the resource of new data sets that they present and with an existing mouse model that enables astrocyte modulation11, provides neuroscientists with powerful tools with which to specifically manipulate and analyse astrocyte genes in adult mice. The questions we outline, and many others, can now be addressed.Footnote 1
Khakh, B. S. & Sofroniew, M. V. Nature Neurosci. 18, 942–952 (2015).
Chai, H. et al. Neuron 95, 531–549 (2017).
Ben Haim, L. & Rowitch, D. H. Nature Rev. Neurosci. 18, 31–41 (2017).
Zhang, Y. & Barres, B. A. Curr. Opin. Neurobiol. 20, 588–594 (2010).
Molofsky, A. V. et al. Nature 509, 189–194 (2014).
Cahoy, J. D. et al. J. Neurosci. 28, 264–278 (2008).
Bazargani, N. & Attwell, D. Nature Neurosci. 19, 182–189 (2016).
Sloan, S. A. & Barres, B. A. Neuron 84, 1112–1115 (2014).
Sanz, E. et al. Proc. Natl Acad. Sci. USA 106, 13939–13944 (2009).
Srinivasan, R. et al. Neuron 92, 1181–1195 (2016).
Francelle, L. et al. Hum. Mol. Genet. 24, 1563–1573 (2015).