A Field Guide to Receptors

Acetylcholine makes a skeletal muscle fibre twitch and slows a heart. It is the same molecule in both cells, so what is different? The difference is the receptor, the membrane protein that binds a signalling molecule (the ligand) and passes the news on. Receptors come in a small number of designs, and the design sets how fast the cell answers and how long the answer lasts. This lesson is a field guide to those designs.

The membrane problem: three ways across, and one way around

Most ligands are water-soluble: neurotransmitters, peptide hormones, growth factors. The plasma membrane is a double sheet of fat, so they stay outside. The cell still has to find out they are there. A receptor can change what happens inside without letting the ligand through, and it has only three ways to do it:

1. Open a pore. The receptor is an ion channel. The ligand binds, the pore opens, and ions flow.
2. Change the shape of a seven-pass protein. One chain crosses the membrane seven times. Binding on the outside rearranges the bundle, and the new shape switches on a partner protein on the inside.
3. Bring single-pass chains together. Each chain crosses the membrane once. The ligand pulls two chains into contact, and the enzyme parts inside meet.

Some ligands avoid the problem. Steroid hormones such as estradiol dissolve in fat and slip through the membrane, so their receptors wait inside the cell. That makes four architectures.

The map: four architectures

Architecture	Where the ligand binds	What binding does	Examples	Time to response
Ligand-gated ion channel	Outside, on a multi-subunit pore	Opens the pore	Nicotinic acetylcholine receptor	About a millisecond or less
Seven-pass (G-protein-coupled)	Outside, in or near the helix bundle	Changes the bundle’s shape, which switches on a G protein	Muscarinic acetylcholine receptor, beta-2 adrenergic receptor	Tens of milliseconds to seconds
Single-pass, enzyme or enzyme-recruiting	Outside, on the outer domain	Brings chains together so kinase domains, or the proteins they recruit, act	EGF receptor, TGF-beta receptors, cytokine receptors, Fas	Seconds to minutes for the first steps; hours for new genes
Nuclear	Inside the cell	Hormone crosses the membrane; receptor holds DNA and recruits gene-reading machinery	Estrogen receptor	Tens of minutes to hours

Ligand-gated channels. The muscle acetylcholine receptor, the nicotinic type, is five subunits arranged around a central pore. Acetylcholine binds on the outer part of the protein, the pore opens, and sodium flows into the muscle cell. The ions are the signal. Seven-pass receptors. These are the G-protein-coupled receptors, or GPCRs. The human genome has about 800 of them, and roughly half detect smells. Binding changes the shape of the helix bundle, and the inner face then acts on a G protein, a switch protein that is off while it holds GDP and on while it holds GTP. Robert Lefkowitz and Brian Kobilka shared the 2012 Nobel Prize in Chemistry for working out how these receptors are built and how they work, and a later lesson is theirs.

Single-pass receptors. About 1,300 human proteins are receptors that cross the membrane once. The part outside the cell is the ectodomain. Receptor tyrosine kinases carry a kinase, an enzyme that attaches phosphate to other proteins, in the part inside the cell, and humans have 58 of them. Receptor serine/threonine kinases, such as the TGF-beta receptors, do the same on a different amino acid. Cytokine receptors have no kinase of their own and recruit kinases from the cytosol, the fluid interior. Death receptors, such as Fas, recruit adaptor proteins that switch on enzymes that take the cell apart.

All of these share one puzzle. Growth factors are often dimeric, built from two identical halves, and receptors are monomeric. How might ligands activate receptors? A dimeric ligand such as NGF can hold two receptor chains at once. A single-chain ligand such as EGF binds one receptor and changes the shape of its ectodomain, exposing a surface that grips a second receptor. Either way, two kinase domains end up close enough to phosphorylate each other. The insulin receptor starts out as a pair, and insulin changes how its two halves sit against each other, which has the same effect.

Nuclear receptors. A nuclear receptor is one chain with two working parts: a pocket that closes around the hormone and a region that grips a specific stretch of DNA. Once the hormone is bound, the receptor holds that DNA and recruits the machinery that reads the gene. There is no second messenger and no kinase cascade, because the protein that received the signal acts directly on the gene that answers. Humans have 48 nuclear receptors, and their ligands include estradiol, thyroid hormone and retinoic acid.

How the map was drawn: sequences first

Until the 1980s a receptor was known by what it bound and what it did. A few had been purified, but the sequence of a whole receptor was still out of reach for most. In 1982 Shosaku Numa’s group in Kyoto changed that. The electric organ of the ray Torpedo is a battery built from modified muscle cells and is packed with acetylcholine receptor. From that organ’s messenger RNA they read the sequence of one receptor subunit, the first complete sequence for a neurotransmitter receptor. Two years later, from the electric eel, they read a voltage-gated sodium channel. Receptors and channels could now be compared with each other.

The next sequences showed families. In 1982 Ora Rosen’s laboratory showed that insulin switches on a tyrosine kinase that travels with the insulin receptor and attaches phosphate to tyrosines on the receptor itself. Axel Ullrich cloned the EGF receptor in 1984 and the insulin receptor in 1985. The EGF receptor resembled erbB, a viral cancer gene, and the insulin receptor’s enzyme part belonged to the same tyrosine kinase family.

Elwood Jensen’s route started with a rat. By 1958 he had seen that the uterus and vagina keep radioactive estradiol, unchanged, while other tissues clear it, so something in those tissues seemed to be catching the hormone. David Toft and Jack Gorski extracted that protein in 1966. In March 1986 two laboratories, Pierre Chambon’s in Strasbourg and Geoffrey Greene’s in Chicago, reported its sequence within weeks of each other. Both noticed that part of it resembled v-erbA, a viral gene derived from a thyroid hormone receptor. The estrogen receptor had relatives.

One ligand, two receptors: acetylcholine and glutamate

Sometimes a ligand can bind two distinct receptor classes: ionotropic versus metabotropic. An ionotropic receptor is itself an ion channel. A metabotropic receptor acts through a G protein and other steps inside the cell.

Acetylcholine uses both. At the muscle end-plate it binds the nicotinic receptor, which is a channel. Colquhoun and Sakmann recorded the current through single receptors at the frog end-plate, and each receptor opened in a burst with brief closures. Binding and response are one event, and nothing has to be made first. In a heart atrial cell, acetylcholine binds the muscarinic receptor, a GPCR. The receptor switches on a G protein, and the G protein opens a potassium channel. The G protein route has a cost. Before a G protein can bind GTP, it must release the GDP it holds. Breitwieser and Szabo estimated, from recordings in heart atrial cells, that unaided the G protein does this only about 0.44 times per minute, a wait of a little over two minutes. A receptor with acetylcholine bound makes the release much faster, and half the maximum current is reached at about 160 nM acetylcholine.

Glutamate, the main excitatory transmitter of the brain, does the same thing. The AMPA and NMDA receptors are channels, and the metabotropic glutamate receptors, the mGluRs, are seven-pass.

Speed follows architecture

A response cannot arrive before its slowest step. The delay is the sum of the steps between the ligand and the effect.

time to response = time to bind + time for each relay step + time for the last step

A channel has no relay step, so its delay is binding plus the opening of the pore, about a millisecond or less. A route with a nucleotide exchange and an enzyme in the middle cannot respond in less than a millisecond, and the G protein route above usually takes tens of milliseconds or more. A single-pass kinase receptor adds pairing, phosphorylation and the recruitment of partners, so the first steps take seconds to minutes and changes in gene expression take hours. A nuclear receptor skips the relay, but the cell still has to make new RNA and protein, which takes tens of minutes to hours.

The time buys two things. The first is amplification. With a protein in the middle, one occupied receptor can switch on many molecules of the next protein, so a short contact outside the cell becomes a large signal inside. The second is control, because the cell can regulate the middle step. A channel that opens when a ligand binds offers little of either: one receptor opens one pore, which passes thousands of ions but makes no copies of an enzyme or messenger, and there is no middle step to adjust. The channel route also ends the moment the transmitter leaves, which is why the gap between nerve and muscle must be cleared quickly. The G protein route lasts until the G protein shuts itself off and the messengers it made are destroyed.

Design problem

Two jobs. First, make sure a heart cell’s next beat is not late. Second, commit a stem cell to a new identity. Choose a receptor architecture for each and justify it by speed, duration and reversibility.

One way to do it

The heart cell: a ligand-gated channel. (This is a design exercise: the heart’s own acetylcholine receptor, described above, is a G-protein-coupled receptor.) The deadline is the time between beats, so the measurement that matters is the delay between ligand and response. A route with a nucleotide exchange and an enzyme in the middle cannot respond in under a millisecond, and a receptor that is itself the channel answers in the same event as binding. The route is also easy to reverse: current flows while the transmitter is bound and stops when it leaves, so the next message can be told apart from the last. The price is no enzymatic amplification and no middle step to regulate, so the transmitter must be released in quantity and cleared fast. If the deadline were a fraction of a second rather than a millisecond, a G-protein-coupled receptor such as the atrial muscarinic receptor could work, and it would give you a step to regulate.

The stem cell: a nuclear receptor. Here speed does not matter. A new identity means a new set of genes switched on, and hours are fine. Duration and reversibility are what matter: a brief pulse of ions would leave no trace, whereas a receptor that sits on the genes starts a program that the new regulatory proteins it makes can keep going after the hormone is gone. A fat-soluble signal such as retinoic acid reaches the receptor directly. A single-pass kinase receptor would also work and is common in fate decisions, but it adds relay steps between ligand and gene.