The Other Brain You Didn’t Know You Had
Astrocyte Glial Cells in the Brain
Imagine, if you will, a recipe you know well. Something delicious you’ve never made yourself but have seen your mom make many times. Let’s assume it’s pumpkin pie. The receipe calls for canned pumpkins, heavy cream, eggs, cornstarch, sugar, maybe even a touch of nutmeg. You’ve seen your mom make it more times than you can count. Geez, you could practically make it yourself from memory. But what if in reality, as it turns out, your mom’s pumpkin pie also requires sardines. Almost as much sardines as pumpkin. And to your dismay, the pie just won’t work, it just won’t taste like pumpkin pie, without it.
In a way, that’s kind of the situation with the brain. Most people know that the brain consists of a huge interconnected network of neurons. More often than not neurons are equated with the term ‘brain cell’, and used interchangeably. But this is not just misleading, it’s completely wrong. Neurons, are of course the critical players that determine how the brain learns and processes information. But they are by all means not the only type of cell in the brain. There are roughly 85 billion neurons in a typical adult brain. But there are also another 86 billion cells in the brain that are not neurons. The main non-neuronal cell type are glial cells. They are the unexpected sardines in our pumpkin pie. There are three major classes of glial cells. Oligodendrocytes speed up electrical signaling in some neurons. Microglia act as the immune cells of the brain and spinal cord, because the normal immune cells and antibodies in your blood don’t have access to the isolated chemical environment of the central nervous system.
We literally have a second network in our brains that plays a critical role in maintaining brain health, but which also affects learning and information processing in ways we do not yet understand.
Astrocytes are the third important class of glial cell. Roughly speaking, there are about as many glial cells in your brain as there are neurons. And about 20–40% of glial cells are astrocytes. It’s the astrocytes I want to focus on here. Without them, neurons would not be able to function properly. But more than that, astrocytes contain the molecular machinery capable of listening in on neuronal signaling, and modulating it. They form a very large network onto themselves, capable of communicating with each other at the same time that they cross-talk with the neuronal network. We literally have a second network in our brains that plays a critical role in maintaining brain health, but which also affects learning and information processing in ways we do not yet understand.
We lack anything close to a coherent mechanistic understanding of their computational role in the brain. This is one of the most exciting areas of neuroscience at the very boundry of our knowledge about the brain.
If you haven’t heard of astrocytes, or only have in passing, you’re in good company, because the truth is that even though we know quite a bit about their molecular biology, biophysics, and chemical makeup, neuroscientists still have no real idea how they interact and influence neuronal signaling and information processing. There are a number of ideas and hypothesis, and accumulating data, but we lack anything close to a coherent mechanistic understanding of their computational role in the brain. This is one of the most exciting areas of neuroscience at the very boundry of our knowledge about the brain.
The German pathologist Rudolf Virchow first described neuroglia in 1858 as a “substance… which lies between the proper nervous parts, holds them together and gives the whole its form in a greater or lesser degree”. The term ‘glia’ has the same origin as the word ‘glue’, since it was initially thought that neuroglia were just there to bind neurons together.
Of course, glial cells do much more than that. In partricular, the classical role of astrocytes is providing homeostatic support for neurons. These cells support and maintain a chemical environment around neurons that allows them to function properly. Because neurons are electrically excitable cells — they communicate between each other by passing signals and messages encoded as fluctuating changes of electric potential across the cell membrane — they are susceptible to becoming overly excited and activated, leading to an inability to function properly. For example, lingering amounts of neurotransmitters, the signaling chemicals that pass messages between neurons, or excess amounts of sodium and potassium ions, the carriers of electric charge in neurons, run the risk of confusing how neurons interpret messages, and can also over-excite them to the point that the cells can die. Astrocytes form a network onto themselves, separate from the neuronal network. In order to maintain a homeostatic environment for neurons, astrocytes ‘mop up’ these excess chemicals and shuttle them internally through their network to blood vessels. The cells that make up the last layer of the astrocyte network send out processes called endfeet that come into close contact with blood vessels. These ‘waste’ chemicals are effectively dumped into the blood stream and whisked away. In the reverse direction, astrocytes shuttle glucose and related energy byproducts from the blood stream to areas of high neuronal activity, where neurons hard at work need energy the most.
But the functional role of astrocytes extends beyond homeostasis. In the early 1990’s neuroscientists recognized that the neurotransmitter glutamate, one of the most important and ubiquitous excitatory neurotransmitters released by neurons, was capable of activating astrocytes and inducing long range signaling within astrocyte networks. In other words, astrocytes, it turns out, have the ability to eavesdrop and listen in as neurons talk to each other. Unlike neurons though, which communicate with each other via electrical excitation, astrocytes communicate via waves of calcium activity. What’s arguably even more amazing is that astrocytes don’t just detect neurotransmitter signaling between neurons, they also have the capability to release neurotransmitters that reciprocally affect how neurons respond. In effect, they have the ability to modulate and shape information processing in the brain. When neurotransmitters are released by astrocytes they are referred to as gliotransmitters in order to differentiate them from the neurotransmitters secreted by neurons. But they are chemically the exact same thing.
The intimate association between neurons and astrocytes is mediated by a concept referred to as the tripartite synapse, or three part synapse. A synapse is the connection point where two neurons meet and pass signals between one another. When a neuron becomes electrically excited a wave of electrical activity propagates down its axon until it reaches the synapse, which in turn triggers the release of neurotransmitters that act as a chemical message. In a tripartite synapse, a part of the astrocyte wraps around the neuronal synapse in order to listen in and participate in the signaling process.
Beyond neuroscience and the biological brain, incorporating astrocytic-like elements into artificial neural networks might have implications for future instantiations of machine learning and artificial intelligence.
And as if this wasn’t enough, a single astrocyte can extend multiple tentacle-like arms to different synapses, so it can participate in many tripartite synapses. One implication of this fact is that an astrocyte may have the potential to functionally connect pairs of neurons that normally would not be connected. It’s possible that synaptic signaling at one tripartite synapse that has an effect on the astrocyte leads to it modulating a different synapse the astroyte is connected to, even though the pair of neurons in the second synapse aren’t actually connected to the first pair. In effect, an astrocyte can ‘short circuit’ the neuronal network.
Admittedly though (again), to what degree astrocytes modulate neuronal signaling and what ultimate affect they have on how the brain processes information — how it learns, or how it influences your decision making — is subject to speculation at this stage. We just don’t understand this aspect of the brain well enough yet. By extension, we have even less understanding of what roles astrocytes might play in complex neurodevelopmental and neuropsychiatric disorders such as autism or schizophrenia. Astrocytes can even modify synaptic plasticity — the ‘strength’ of the connections between neurons.
Beyond neuroscience and the biological brain, incorporating astrocytic-like elements into artificial neural networks might have implications for future instantiations of machine learning and artificial intelligence. In fact, there are some early indications in attempts at building neuronal-astrocyte-like artificial neural networks that suggest an improved performance on classification tasks. Machine learning and artificial intelligence in general have much to learn from the biological brain.
The brain has a long history of challenging neuroscientists in their attempts at figuring out how it works, and how it fails in diseases. Nowhere is this more true then in our attempts at understanding what the humble and quiet astrocyte is actually doing.