Brain Cells: Neurons and Glia

Two classes of cell

Open any histology slide of brain tissue and you will find a dense, tangled mass of cells packed so tightly that the spaces between them are measured in tens of nanometres. Sort those cells and, with a few specialised exceptions, they fall into two camps. In the first camp are the neurons: elongated, branched, electrically excitable cells that generate impulses and pass them on at synapses. In the second are the glia, a diverse group of non-excitable cells that surround neurons, wrap their axons, regulate the fluid around them, and clean up after them.

The distinction is real and it is functional, not merely descriptive. A neuron can fire an action potential; a glial cell, in the classic sense, cannot. Neurons carry the traffic of information; glia build and maintain the road. But the analogy breaks down quickly, because the road in this case adjusts its own surface, decides which vehicles get through, and occasionally rearranges the junctions. Modern work has shown that glia respond to neuronal activity with waves of calcium, release their own signalling molecules, and change the strength of synapses. They are participants, not scenery.

There is a third population worth naming, because it does not fit either camp cleanly: the vascular cells. Endothelial cells lining the brain's capillaries, together with the pericytes wrapped around them, form the physical wall of the blood-brain barrier. They are not neurons and they are not classical glia, but no account of the brain's cells is complete without them, because they decide what molecules ever reach the other two.

How many of each

For decades the standard figures were a round 100 billion neurons and roughly ten times as many glia. Both numbers were repeated so often that they hardened into textbook fact, and neither was ever traced to a rigorous count. The problem was methodological: the brain is not uniform, so counting cells in a small sample and multiplying up gives an answer that depends heavily on which sample you chose. Cortex and cerebellum differ enormously in cell density, and any estimate that ignores that will be wrong.

The situation changed with the isotropic fractionator, a technique that dissolves a whole brain region into a homogeneous suspension of free nuclei, stains the neuronal nuclei with a specific marker, and counts them directly. Applied to whole human brains, the method gave a figure of approximately 86 billion neurons and a similar order of non-neuronal cells, with an overall ratio close to one to one rather than ten to one.

The regional detail is as interesting as the total. The cerebellum, a structure that occupies a small fraction of brain volume, contains the great majority of all the brain's neurons, because its granule cells are exceptionally small and densely packed and its glia-to-neuron ratio is low. The cerebral cortex, by contrast, holds a minority of the neurons but a large share of the glia, so within the cortex alone glia genuinely do outnumber neurons. The old ten-to-one claim was, in effect, a cortical observation wrongly generalised to the whole brain.

What makes a neuron distinctive

Every cell in the body has a membrane, a nucleus, and metabolic machinery. What sets the neuron apart is a combination of three properties, and it is the combination, not any one of them alone, that makes signalling possible.

Take any one of these away and the cell stops being a neuron in any useful sense. A cell that is excitable but not polarised cannot direct a signal. A cell that is polarised but not connected has nowhere to send it. A cell that is connected but not excitable can still communicate, and indeed astrocytes do, but slowly, chemically, and diffusely rather than at the millisecond precision that neural computation requires.

The glial families

Glia are not one cell type but four, and they have almost nothing in common with each other beyond not being neurons. Each family has a distinct developmental origin, a distinct morphology, and a distinct job. Lumping them together under one word is a historical accident that continues to mislead.

The most consequential correction of the last few decades concerns astrocytes. They were long thought to be metabolic housekeepers, and they are, but they also sit directly at the synapse. An astrocytic process wraps around the presynaptic terminal and the postsynaptic membrane, senses the transmitter released between them, and responds by releasing signalling molecules of its own that alter how strong the connection is. A synapse, on this view, is not a two-part junction at all.

The blood-brain barrier

The brain runs on a chemistry so tightly specified that even modest fluctuations in the ionic composition of its extracellular fluid would disrupt signalling. Blood, meanwhile, is a chemically turbulent medium: its concentrations of glucose, amino acids, hormones, and ions swing with every meal, every exertion, and every stress response. The brain solves this incompatibility by refusing to let the two mix freely.

The blood-brain barrier is the structure that enforces the separation. Unlike capillaries elsewhere in the body, which are leaky by design, the endothelial cells lining brain capillaries are stitched together by tight junctions that seal the gaps between them. Nothing crosses by simply slipping between cells. To enter brain tissue, a molecule must either be small and fat-soluble enough to diffuse straight through the cell membranes, as oxygen, carbon dioxide, and many anaesthetics do, or it must be recognised by a specific transporter protein, as glucose and certain amino acids are.

Astrocytes are integral to this arrangement. Their end-feet cover the outer surface of the capillaries and signal to the endothelium, helping to induce and maintain the tight junctions. This is a case where a glial cell is not supporting neurons directly but constructing the physical infrastructure on which all neuronal function depends. It is also the reason so many drugs that work beautifully in a test tube do nothing in a living brain: they never get in.

How the cells work together

The useful mental model is not neurons plus a support crew, but a single integrated tissue in which several cell types each hold part of the job. Consider what has to be true for one synapse to fire reliably, thousands of times a second, for eighty years.

1. Fuel has to arrive

   The brain is metabolically ravenous, consuming roughly a fifth of the body's energy while making up a fiftieth of its mass. Glucose crosses the barrier through dedicated transporters and is taken up by both neurons and astrocytes. On a widely discussed but still-debated model, astrocytes also pass lactate to neurons as a supplementary fuel during bursts of activity; how much of the brain's working energy actually travels by that route is contested. What is not in doubt is that the supply chain is unforgiving: interrupt it, and signalling stops within seconds.

2. The chemistry has to stay clean

   Every action potential dumps potassium into the extracellular space, and every synaptic release leaves glutamate behind. Left to accumulate, both would be catastrophic: excess glutamate is excitotoxic and kills neurons outright. Astrocytes take up both, continuously, and convert glutamate back into a harmless precursor.

3. The wiring has to be insulated

   Oligodendrocytes wrap axons in myelin, raising conduction speed from around one metre per second to well over a hundred. The brain's ability to coordinate distant regions on a millisecond timescale depends entirely on this glial contribution.

4. Damage has to be cleared

   Microglia patrol the tissue continuously, extending and retracting processes to sample their surroundings, engulfing debris, dead cells, and pathogens. During development they also prune weak or surplus synapses, sculpting circuits by subtraction.

5. The synapse has to be tuned

   At the tripartite synapse, the astrocyte listens to the transmitter passing between neurons and adjusts the connection accordingly, sharpening it, damping it, or clearing it faster. The strength of a connection is therefore set by three cells, not two.

Seen this way, the old division into signalling cells and support cells is not so much wrong as misleadingly ranked. Neurons do the computing, but they can only compute because a second population is continuously maintaining the physical and chemical conditions under which computation is possible, and is quietly modulating the computation while it does so.

Explore brain cells

Related topics