2 Unit III — Signaling

Overview

The previous unit left us with a body that needs governing and a control system built to govern it. In the chapter on the hypothalamus we watched that control system reach out into the body and the body signal back: circulating molecules in the blood reporting the state of the periphery to the brain, and hormones released by the brain acting on distant tissues — and, returning, acting on neurons themselves. This unit is about the mechanisms underneath all of that. The word for the whole business is signaling, and the argument of the unit is contained in a single observation: signaling happens across an enormous range of timescales, from thousandths of a second to days, and the neuron is a participant at every scale — but for different reasons at each.

That observation is not original to me. The neuroscientist Robert Sapolsky has made a habit of asking, about any piece of behavior, on what timescale do its causes operate? — what happened a second before, an hour before, a lifetime before, an evolutionary epoch before. The same discipline applies inside the single neuron. A neuron responds to a glutamate molecule in a millisecond and to a steroid hormone over hours, and these are not the same kind of event wearing different clothes. They are different mechanisms, serving different purposes, operating at different speeds, and a neuron is built to live in all of them at once. Most introductions to neural signaling collapse this range down to its fastest layer — the firing cell, the spike, the synapse — and treat everything slower as background. I want to resist that from the start, because the fast electrical layer is not the foundation of signaling. It is the most recent, most specialized, and by a wide margin most expensive part of a much older system.

2.1 Why signal at all

Begin with the reason a cell would need to signal another at all. The first unit’s answer was movement: a nervous system exists to coordinate the activity of a body so that the organism can move adaptively — toward food, away from threat, toward a mate. Even the simplest coordinated movement requires that one part of the system communicate with another. A circuit that alternates a limb forward and back must have some cells exciting the muscles that flex and others inhibiting them, in turn, in time. Coordination is communication, and communication of exactly two basic messages — excite and inhibit — is the raw material from which all of it is built. Everything in this unit is ultimately machinery for sending those two messages, in the right place, at the right time, and on the right timescale.

2.2 Forms of signaling

Cells signal one another in several physically distinct ways. The differences come down to how far the message travels and how it gets there, and the four basic forms run from the most far-reaching to the most intimate:

Endocrine signaling sends a molecule — a hormone — into the bloodstream, to be carried throughout the body and act on any cell bearing the right receptor. It is slow, because it depends on circulation, and broad, because the blood goes everywhere.
Paracrine signaling is local: a molecule is released and diffuses a short distance to neighboring cells. Synaptic transmission, as we will see, is essentially a tightly controlled form of paracrine signaling.
Autocrine signaling is a cell signaling itself — its own released molecule binding receptors on the very cell that released it.
Direct signaling dispenses with the released molecule altogether. Through gap junctions, the interiors of two adjacent cells are physically continuous, and ions or small molecules pass straight from one to the other. This is the basis of the electrical synapse, the fastest form of communication between cells.

A point that will recur throughout the unit is already visible in these categories. The systems of the body are named for their messengers — hormones in the endocrine system, cytokines in the immune system, neurotransmitters and neuromodulators in the nervous system — but the messengers do not respect the boundaries. Many of the very same molecules serve in more than one system. This is the first hint of a principle we will develop carefully: what a signaling molecule means is not a fixed property of the molecule. We have, in fact, already seen this principle at work. In the previous unit, circulating hormones reported the body’s state to the brain and acted directly on neurons. The molecules that did so were not built for the brain; they were endocrine signals that neurons happened to be equipped to read.

2.3 Where the vocabulary came from

Before going further into the machinery, it is worth a brief digression on how old this vocabulary is — because it is far older than the nervous system itself.

Consider the marine sponge. It is about as simple as an animal gets: no neurons, no nervous system, no electrical transmission of the kind we are about to study. It is a filter feeder, drawing water through its channels to strain out bacteria. Yet it faces a control problem, because those channels clog, and when they do the animal clears them with a slow whole-body contraction that looks, reasonably enough, like a sneeze. The sneeze must be coordinated, and the sponge has no nerves to coordinate it. What it has is chemistry. The contraction depends on the release of glutamate, and it can be suppressed by GABA — the very molecules that serve as the principal excitatory and inhibitory transmitters of your own brain. The animal with no nervous system is already running its coordination on the chemical signals you are using to read this sentence.

The molecules of signaling, in other words, are inherited, not invented. The nervous system did not design its vocabulary from scratch; it specialized in the use of molecules that cells had been signaling with for a very long time. And we can say something about where those molecules came from. Every cell is a web of chemical reactions, and the concentration of any reaction’s product is itself a readout of how fast that reaction has been running. A molecule sitting at some concentration is therefore already carrying information about the state of the machinery that makes it — already, in a minimal sense, a signal. Evolution did not have to invent messengers; it had only to repurpose molecules that were already reporting on the cell’s internal state.

2.4 Membranes and the cost of a mobile cell

That is enough about the molecules. The deeper question for this unit is not where the signals came from but why signaling has the character it does — and in particular why its fastest layer is so costly. To answer that we have to start not with neurons but with the boundary that makes a cell a cell: the membrane.

Without a membrane there is no cell — nothing to distinguish a single-celled organism like an amoeba from the pond water around it. The membrane is the partition that separates life from its environment. But the moment a cell wraps itself in a membrane, it has a problem, because a perfectly sealed membrane would be a tomb. A cell cut off from its surroundings could not respire, take in nutrients, or expel waste. A membrane therefore cannot be a perfect barrier; it must be permeable, punctuated by pores and channels and receptors that allow the regulated traffic of matter and information. Permeability is not optional. It is forced on the cell by the need to stay alive.

And permeability immediately creates a vulnerability. A cell is full of dissolved solutes, so its interior is more concentrated than fresh water. Put such a cell in pond water and the concentration gradient will drive water in across the permeable membrane, swelling the cell and threatening to burst it. (This is why dried beans, soaked overnight, come out swollen the next morning.) This osmotic threat is a genuine engineering crisis, and evolution solved it in two very different ways. Plants solved it structurally, encasing each cell in a thick rigid wall that resists the pressure. The solution works, but it has a cost: a cell in a rigid box cannot move. Plants are sessile, and must make their own food where they stand because they cannot go and find it.

Animals took the other path. An animal makes its living by moving — to forage, to escape, to find a mate — and movement is incompatible with a rigid wall. So the animal cell keeps a thin, flexible membrane, and pays for that flexibility by managing the osmotic threat actively rather than structurally. It continuously pumps ions across its membrane, manipulating the electrochemical gradients across the boundary to keep its interior from flooding. Pumping ions against their gradients is uphill work — like pumping water up into a water tower against gravity — and uphill work costs energy. The animal cell, having given up the wall for the freedom to move, must spend metabolic energy continuously, for its whole life, simply to hold its own contents in balance.

Here is the connection that matters for this unit, and it is worth stating carefully, because it is easy to overstate. Those pumped ion gradients — a great deal of sodium held outside the cell, a great deal of potassium held in — are a store of potential energy maintained across the membrane. The neuron’s fast electrical signals are, mechanically, the controlled discharge of that store: a channel opens, ions rush down the gradient, and the membrane voltage changes. The machinery that an animal cell evolved to survive without a wall is the same machinery the neuron would later exploit to signal quickly. I do not mean that the action potential is osmotic regulation, or that managing osmolarity is the origin of neural communication — that would be a tidy overstatement of a real continuity. The honest version is more modest and more interesting: fast electrical signaling did not require a new energy source. It spends a store of energy that animal cells were already maintaining for an older reason, and it inherits that store’s continuous metabolic price.

That price is enormous. The human brain is about two percent of body weight and consumes roughly twenty percent of the body’s energy at rest. A large share of that goes to a single molecular machine — the sodium–potassium pump — doing nothing more glamorous than maintaining the disequilibrium that keeps neurons poised to fire. We asked, at the end of the first unit, what it costs to buy prediction. Here is part of the answer, in metabolism: the fast end of signaling is fast precisely because it is expensive, and the brain pays that bill continuously, for as long as it is alive.

2.5 How a signaling molecule reaches a neuron

With the membrane in view, we can ask the practical question the rest of the unit depends on: when a signaling molecule arrives at a neuron, how does it actually change what the neuron does? There are a few routes, and they differ enormously in speed.

Some molecules simply diffuse straight through the membrane. The lipid bilayer is permeable to small uncharged and fat-soluble molecules, and a few signals — the steroid hormones among them — can cross it directly and act on receptors inside the cell, in the cytosol. Other molecules cannot cross, and must be admitted by a dedicated transporter that ferries them in. The rest act at the membrane itself, by way of channels and surface receptors — and because these are the workhorses of fast neural signaling, they are worth naming and distinguishing now.

The plainest channels are the leak channels, which are simply always open. A leak channel is like a narrow alleyway between two buildings: nothing decides whether it is open, because it is always open — but size still matters, and a molecule too large for the gap does not get through. Leak channels are how the membrane maintains a baseline permeability to particular ions, and we will see that they are central to setting a neuron’s resting state.

Other channels are gated — normally shut, and opened only on cue — and the nature of the cue is what matters. An ionotropic receptor is a channel with a lock. It is like that same alleyway, but with a locked gate across it: the signaling molecule is the key, and when it binds, the gate swings open and ions pass through directly. Because binding is opening, the effect is nearly immediate — a change in the cell’s voltage within a millisecond or two. This is the machinery of fast, precise, point-to-point signaling, and glutamate and GABA both have ionotropic receptors of exactly this kind.

A metabotropic receptor works differently, and more slowly. It is like a doorbell. The signaling molecule presses the bell from outside, but nothing passes through the wall directly; instead the occupants inside are roused into activity, and they decide what happens next. Sometimes the result is that a channel elsewhere on the cell is opened — the occupants come and open a door. Sometimes the result is something else entirely, an errand run inside the house that never opens a door at all. Either way the molecule outside never enters; it only triggers internal machinery, and that indirection takes time. Metabotropic effects unfold over tens to hundreds of milliseconds, and can last far longer than the signal that started them. (We will leave the inner machinery — the G-proteins and second-messenger cascades that the doorbell sets running — for the chapter on neuromodulation, where it does its most important work. For now, what matters is the timing: a key opens a gate at once; a doorbell sets off a slower commotion whose outcome varies.)

The same fork appears one more way, and it returns us to the membrane-crossing molecules we began with. A surface receptor that triggers an internal cascade need not open a channel at all — it may reach inward toward the cell’s deeper machinery, even toward the genome. And some molecules skip the surface entirely and act on receptors waiting in the cytosol. The consequences of a signal, then, range from a gate opening in a millisecond to a change in gene expression unfolding over hours — and which of these happens is determined not by the molecule but by the receptor it meets. No single molecule shows this more vividly than a steroid hormone such as cortisol. Cortisol is fat-soluble, so it can diffuse through the neuron’s membrane and bind receptors waiting in the cytosol; the resulting complex travels to the nucleus and changes which genes the cell expresses — a slow, deep, long-lasting effect unfolding over hours. But the same cortisol molecule can also bind a receptor on the membrane’s outer surface and trigger a rapid second-messenger cascade that alters the neuron’s activity within seconds. One molecule, two receptors, two mechanisms, two timescales — the fast one measured in seconds, the slow one reaching into the genome and lasting days. The molecule does not carry its meaning with it. Its meaning is decided where it lands. (The same is true of the other steroid hormones, testosterone and estrogen among them, which likewise cross the membrane to act on the genome while also acting at the surface — a fact that will matter when we come to how these molecules shape the brain.)

There is one more dimension to this, and it is subtle enough to be worth isolating, because it turns concentration itself into a kind of message. A given molecule is often read by several receptor types that differ in their affinity — in how tightly they bind it, and therefore in the concentration at which they respond. A high-affinity receptor is occupied even when the molecule is scarce; it is sensitive to the molecule’s baseline, tonic level. A low-affinity receptor stays empty until the molecule becomes abundant; it responds only to a surge, a phasic burst. Run a single molecule past both, and the cell can tell the difference between “a little, all the time” and “a lot, just now” — because the two conditions engage different receptors and so produce different effects. Cortisol again provides the clean example: its high-affinity mineralocorticoid receptors are largely occupied at the body’s resting cortisol level and track normal daily regulation, while its lower-affinity glucocorticoid receptors are recruited only when cortisol spikes under stress, switching on a distinct genetic program. The hormone says one thing at rest and another under stress, using the same molecule, by means of receptors tuned to different concentrations. Affinity makes concentration an information channel in its own right.

2.6 The argument, and the unit ahead

Step back and the shape of the argument is the timescale itself. An electrical synapse passes a signal between coupled cells almost instantaneously. A glutamate molecule at an ionotropic receptor changes a neuron’s voltage in a millisecond or two. A neuromodulator working through a surface receptor and an internal cascade shifts the neuron’s excitability over seconds. A steroid hormone reaching the genome remakes the neuron over hours and leaves a trace for days. A hormone circulating in the blood sets the body’s operating state over still longer spans. These are not five versions of one thing. They are genuinely different mechanisms, and the neuron takes part in all of them — using the fast ones to carry information moment to moment, and the slow ones to regulate how that information is carried and whether it leaves a lasting mark. This is the sense in which, following Sapolsky, we will treat the neuron as a creature of many timescales at once.

The chapters ahead descend through these timescales in turn. We begin with the cellular cast — neurons, glia, and the synapse between them — because we should know the players before we watch them act. We then build the fast electrical layer from first principles: membranes and gradients in detail, the channels that open and close, the way a neuron sums its inputs, and the action potential that carries the result down the axon — along with the metabolic price we have just previewed. From there we widen into the slower chemical environment in which all that fast signaling is bathed: neuromodulation, the diversity of receptors, and the difference between a signal aimed at a single synapse and one released to drift across many cells at once. We give the glia their own chapter, and with them the broader traffic between the nervous system and the body’s other signaling systems, because the synapse is not the only place signaling happens. And we end where signaling becomes memory — the activity-dependent changes that let a synapse record its own history, how those changes are reinforced, and a first look forward at how a brain might learn to assign credit for a reward that arrives only after the act that earned it.

Throughout, the molecule stays the humble inherited thing it was in the sponge. What grows — what made a nervous system possible, and what organizes everything ahead — is the range of timescales over which a neuron can be made to respond, and the receptors that decide, in each case, what a simple chemical signal is going to mean.