Cell Signaling · Unit 3: G-protein-coupled receptors and second messengers

GPCRs and Trimeric G Proteins

How can one receptor fold report adrenaline, acetylcholine, cannabinoids and light, and how can one hormone give opposite answers in two tissues? This lesson shows what an agonist changes in a seven-pass receptor, how the receptor loads a trimeric G protein with GTP, and what the four Gα families and the βγ pair do next. Afterwards you can predict a tissue's first response from its receptor subtype, and tell an agonist, an antagonist and an inverse agonist apart.

Watch first The Receptor Decides the Answer · 2:55

Adrenaline makes your heart beat harder, relaxes the muscle around your airways and narrows the blood vessels in your skin. How does one molecule give each tissue a different order? The hormone carries no instruction about the answer. Each cell decides, through the receptor subtype it puts on its surface and the G protein that receptor hands the signal to. This lesson covers that receptor family, the G-protein-coupled receptors (GPCRs), and the three-part G proteins they switch on.

The idea: one fold, lots of inputs

A GPCR is a single protein chain that crosses the plasma membrane seven times. Each crossing is an alpha helix of twenty or more water-avoiding amino acids. The amino end of the chain lies outside the cell and the carboxyl end inside. The seven helices pack into a bundle with a pocket open to the outside and loops facing the cytosol. You met this architecture in the receptor field guide.

“Lots of inputs/ligands.” Receptors built on this one plan detect adrenaline, acetylcholine, dopamine, histamine, peptide hormones such as glucagon, odorants, bitter and sweet tastes, and light. The human genome holds roughly 800 GPCR genes (Lefkowitz’s Nobel lecture says almost a thousand), about half of them for odorant receptors, and a mouse has about a thousand odorant receptor genes alone. Roughly a third of approved drugs act on a GPCR.

The prototype is a protein that catches light. Rhodopsin, the pigment of the rod cells in the retina, was too greasy to sequence whole. Two laboratories, Paul Hargrave’s in the United States and Yuri Ovchinnikov’s in Moscow, cut it into pieces with a chemical that breaks the chain only at methionine and sequenced the pieces one amino acid at a time. By 1983 both had published the complete sequence of cow rhodopsin. It held seven runs of water-avoiding residues, each long enough to span a membrane.

Three years later the same plan turned up in an unexpected place. In 1986 the laboratory of Robert Lefkowitz at Duke, working with a group at Merck, cloned the gene for a β-adrenergic receptor, a protein adrenaline acts on. Brian Kobilka, a cardiology fellow in the laboratory, had worked toward the gene for almost two years. When the new sequence was compared with known proteins, it resembled rhodopsin, seven spans included. The only other seven-pass protein then known also worked with light, and Lefkowitz recalled in his Nobel lecture that it had been speculated that seven spans were a signature of light-sensitive molecules. Adrenaline has nothing to do with light. The seven spans had to mark something else: the family of receptors that pass their signal to a G protein.

What the agonist changes, and what the receptor does next

An agonist is a ligand that activates its receptor, as in the lesson on binding and dose-response. Adrenaline binds in the pocket among the helices, toward the outer face of the membrane. The contacts it makes shift the helices a little on the outside and a lot on the inside. Compare the crystal structure of the inactive β2-adrenergic receptor with the one caught holding its G protein: the cytoplasmic end of helix 6 has swung outward by about 14 Å. That opens a cavity on the inner face of the receptor, and the tail of the G protein’s alpha subunit fits into it.

The G protein waiting there has three subunits. Gα holds a guanine nucleotide. Gβ and Gγ bind each other so tightly that they act as one unit, written βγ. Lipid tails on Gα and Gγ keep the trio on the inner face of the membrane, as in the lesson on lipid anchors. While Gα holds GDP the three stay together and are silent.

The active receptor is the exchange factor (the GEF) from the lesson on GTPase switches. It grips the G protein and loosens Gα’s hold on GDP. GDP leaves, and GTP, which is more plentiful in the cytosol, takes the empty site. In the textbook picture Gα-GTP then lets go of both the receptor and βγ, though measurements in living cells suggest that Gi may rearrange with βγ rather than fully separate.

active receptor + Gα(GDP)·βγ + GTP → active receptor + Gα(GTP) + βγ + GDP

The receptor appears on both sides because it is a catalyst. Still holding its agonist, it can load another G protein, which is one source of the gain you saw in the lesson on signaling circuits. The cell now has two messengers, Gα-GTP and free βγ, and each can find a target. The signal ends when Gα hydrolyses its GTP and takes βγ back.

”All the pathways”: four Gα families and βγ

Hundreds of receptors feed a small number of G proteins. The human genome has 16 genes for Gα, and by sequence they fall into four families. Learn the main target of each and you can predict the first step after almost any GPCR.

Gα familyMain direct targetFirst consequenceReceptors that use it
Gsadenylyl cyclase, switched oncAMP risesβ-adrenergic, glucagon
Gi/oadenylyl cyclase, held backcAMP falls; freed βγ actsM2 muscarinic, α2-adrenergic, CB1 cannabinoid
Gq/11phospholipase C-βPIP2 is cut into IP3 and DAG; calcium risesα1-adrenergic, M3 muscarinic
G12/13exchange factors for Rhoactin is rearranged; cells change shape and movethrombin, lysophosphatidic acid

βγ is a signal in its own right. Its best-known job is in the heart, where βγ released from Gi binds a potassium channel called GIRK (G-protein-gated inwardly rectifying K+ channel) and opens it.

“Lots of downstream effectors.” Picture an hourglass: many receptors narrow to four Gα families plus βγ, and these fan out again to enzymes and ion channels.

How do you find out which family a receptor uses? One classic tool is pertussis toxin, made by the bacterium that causes whooping cough. It is an enzyme that attaches ADP-ribose to a cysteine near the tail of Gαi and Gαo, the part that fits into the receptor. A G protein modified this way cannot be activated by its receptor, so it stays in the GDP form and its pathway goes quiet. Cholera toxin, from the lesson on GTPase switches, does the reverse to Gs and jams it on. If pertussis toxin abolishes a response, that response runs through Gi/o.

The test helped give a name to an orphan, a receptor gene with no known ligand. In 1990 Lisa Matsuda and colleagues, in Tom Bonner’s group at the National Institute of Mental Health, had a rat brain gene that predicted a seven-pass protein. Allyn Howlett’s laboratory had already shown in nerve-cell preparations that cannabinoids lower cAMP production in a dose-dependent way, that only one mirror-image form of the drug works, and that pertussis toxin blocks the effect. Matsuda expressed the orphan in cultured cells that lacked it, added cannabinoids, and reported the same properties. The orphan was the cannabinoid receptor now called CB1, and it couples to Gi/o.

Same ligand, opposite answers

The receptor subtype and the G protein it couples to decide the answer, not the hormone.

Take acetylcholine. Pacemaker cells of the heart carry the M2 muscarinic receptor, which couples to Gi. The freed βγ opens GIRK channels, potassium flows out, the membrane potential moves away from the firing threshold, and the heart slows. Smooth muscle around the airways carries the M3 muscarinic receptor, which couples to Gq. Phospholipase C-β cuts PIP2, IP3 releases stored calcium, and the muscle contracts. One transmitter quiets one tissue and excites another. At skeletal muscle acetylcholine meets a different kind of receptor again, an ion channel from the receptor field guide.

Adrenaline does the same with a larger cast. Humans have nine adrenergic receptor genes: three α1 subtypes that couple to Gq, three α2 subtypes that couple to Gi, and three β subtypes that couple to Gs.

Design problem

One hormone must make the heart beat harder, relax the muscle of the airways, and narrow blood vessels in the skin. Assign a receptor subtype and a Gα family to each tissue so that this works. Then predict what a drug that blocks only one of your subtypes would do.

One way to do it

Start with the two smooth muscles, because they must do opposite things. Smooth muscle contracts when cytosolic calcium rises, so the skin vessels get a Gq-coupled receptor: α1. Airway muscle gets a receptor that raises cAMP, which in smooth muscle favours relaxation: β2 on Gs. The heart gets β1, also on Gs. Here the same G protein and the same messenger make muscle contract harder, because heart cells and airway cells have different proteins waiting downstream of cAMP (the next lesson follows that messenger). The receptor and its G protein choose the messenger. The cell’s own proteins decide what the messenger means.

Now block one subtype. A drug that blocks only β1 lowers the rate and force of the heartbeat when adrenaline is present and leaves airways and skin vessels alone. That is the reasoning behind β1-selective beta-blockers. A blocker that also hits β2 removes adrenaline’s relaxing effect on the airways, which is a risk for a person with asthma. This assignment is the one the body mainly uses.

Drugs at GPCRs, and why “GPCRs can be unfaithful to trimeric G proteins”

Watch Three Ways a Drug Moves a Receptor · 1:30

A GPCR is flexible. Even with an empty pocket it flickers between inactive and active shapes, and for many receptors a small fraction is active at any moment. Such a receptor signals a little with no ligand at all. That basal activity gives a drug three ways to act where the lesson on binding gave you two.

  • An agonist binds the active shape more tightly, so it holds receptors there and signaling rises above the basal level.
  • A neutral antagonist binds both shapes equally well. Alone it changes nothing, but it occupies the pocket and keeps other ligands out.
  • An inverse agonist binds the inactive shape more tightly, so it pulls receptors out of the active state and signaling falls below basal.

In a cell with no basal activity an antagonist and an inverse agonist look the same. You need a system with measurable basal activity to tell them apart, and many drugs long called antagonists proved to be inverse agonists when tested that way. Carazolol, the beta-blocker bound in the 2007 crystal structures of the β2 receptor, is a partial inverse agonist.

The table above gives each receptor one partner, and real receptors are less loyal. One receptor can couple to more than one Gα family. The thrombin receptor on platelets activates both Gq and G12/13, so a single signal both raises calcium and reshapes the cell. Which partner dominates can differ between cell types, so a coupling measured in one cell line is a hypothesis about the next. A receptor that has been active for a while also takes a partner of another kind. Kinases phosphorylate its tail and a protein called arrestin binds, which blocks the G protein and assembles a second set of signaling proteins on the receptor. The lesson on turning signals off takes that up.

How we measure it

Cyanogen bromide cleavage and Edman sequencing

Cyanogen bromide cuts a protein chain at methionine, and Edman chemistry then removes and identifies one amino acid at a time from the end of each fragment. Overlapping fragments are pieced into the full sequence. This is how rhodopsin was read before its gene was cloned.

Hydropathy analysis

Score each amino acid in a sequence for how well it tolerates water, then average the scores along the chain. A run of about twenty water-avoiding residues is long enough to cross a membrane as a helix. Seven such runs in rhodopsin and in the β-adrenergic receptor gave the seven-pass model years before a crystal structure of either existed.

Expression of a cloned receptor in cells that lack it

Put the gene for a candidate receptor into a cultured cell line that does not make it, add test ligands, and measure a second messenger such as cAMP. A response that appears only in cells carrying the gene, rises with dose, and is lost after pertussis toxin identifies both the ligand and the G protein family.

Crystallography of stabilised receptors

GPCRs are scarce and unstable in detergent, so crystals were obtained by holding the receptor in one shape: an inverse agonist in the pocket with T4 lysozyme fused in for the inactive state, an agonist with the G protein and a llama antibody fragment for the active one. Comparing the two structures shows what an agonist changes.

Finish line

Test yourself

8 questions

Practice and MCAT-style, with explained answers.

Next · Lesson 13

cAMP and Protein Kinase A

How can one cell with one messenger give two different responses?

Continue →

← Lesson 11 · Lipids as Anchors and Addresses

β2-adrenergic receptor bound to Gs — the active receptor holding its G protein
β2-adrenergic receptor bound to Gs — the active receptor holding its G protein ● LIVE · RCSB PDB 3SN6
open on RCSB PDB ↗
Loading live structure 3SN6 from RCSB PDB…

On the cytoplasmic side, find helix 6 swung outward and the last helix of Gαs sitting in the cavity it opened. The extra proteins are T4 lysozyme and a nanobody used to steady the complex.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
β2-adrenergic receptor, inactive — the same receptor with an inverse agonist bound
β2-adrenergic receptor, inactive — the same receptor with an inverse agonist bound ● LIVE · RCSB PDB 2RH1
open on RCSB PDB ↗
Loading live structure 2RH1 from RCSB PDB…

Carazolol sits in the pocket among the helices, toward the outer face. Compare the inner end of helix 6 with 3SN6: here it is closed against the bundle. The extra domain is T4 lysozyme.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Bovine rhodopsin — the prototype of the seven-pass fold
Bovine rhodopsin — the prototype of the seven-pass fold ● LIVE · RCSB PDB 1F88
open on RCSB PDB ↗
Loading live structure 1F88 from RCSB PDB…

Count the seven helices and find retinal held among them. This crystal structure confirmed the seven membrane crossings that had been inferred from the sequence in the early 1980s.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Heterotrimeric G protein (Gi) with GDP — the resting switch
Heterotrimeric G protein (Gi) with GDP — the resting switch ● LIVE · RCSB PDB 1GP2
open on RCSB PDB ↗
Loading live structure 1GP2 from RCSB PDB…

Gα with GDP bound sits against Gβ, and Gγ runs along Gβ. This is the silent, assembled form that an active receptor acts on.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Adrenaline (epinephrine) — one hormone, nine receptors
Adrenaline (epinephrine) — one hormone, nine receptors ● LIVE · PubChem CID 5816
open on PubChem ↗
Loading live structure 5816 from PubChem…

A catechol ring and an amine on a short chain. The same small molecule fits the pockets of α1, α2 and β receptors, which couple to different G proteins.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Robert Lefkowitz

Robert Lefkowitz b. 1943

His laboratory, with a group at Merck, cloned a β-adrenergic receptor gene in 1986 and found in its sequence the seven membrane spans of rhodopsin, which tied hormone receptors and the visual pigment into one family.

The overview lesson on signal transduction tells the longer story of Lefkowitz and Kobilka. This part of it starts in 1986, when his laboratory at Duke and a group at Merck had cloned the gene for a β-adrenergic receptor and read off its amino acid sequence. Compared with known proteins, the new sequence matched rhodopsin, the light-catching pigment of the eye: both chains had seven stretches able to cross a membrane. In his Nobel lecture Lefkowitz recalled that the other seven-pass protein then known was also light-sensitive, and that it had been speculated that seven spans were a signature of such molecules. A hormone receptor had no reason to look like that. The match meant that the seven spans marked receptors that signal through G proteins, whatever they detect. In the years that followed, his laboratory and others found further receptor genes by their resemblance to those already cloned, and the family that emerged is the target of about a third of approved drugs.

✓ verbatim Lefkowitz, Nobel Lecture, 8 December 2012; Dixon, Kobilka et al., Nature 321:75 (1986).
Brian Kobilka

Brian Kobilka b. 1955

As a fellow in the Lefkowitz laboratory he worked on cloning the β-adrenergic receptor gene. His own laboratory later solved the β2 receptor's structure, inactive in 2007 and bound to Gs in 2011, which showed what an agonist changes.

Kobilka joined the Lefkowitz laboratory as a cardiology fellow and joined the effort to clone the receptor for adrenaline. By his own account for the Nobel Foundation it went slowly. Libraries made from messenger RNA gave only false leads, so the group turned to a library of genomic DNA, and after almost two years the gene came out. Along the way he set up a molecular biology bench inside the laboratory, and Tong Sun Kobilka, his wife, was hired part time to help. Having the gene did not give a picture of the protein. The receptor is scarce and unstable in detergent, and a high-resolution crystal structure came only in 2007, twenty-one years after the cloning, with the phage enzyme T4 lysozyme fused in to help crystals form. In 2011 his group at Stanford caught the receptor holding its G protein, steadied by an antibody fragment from a llama. Comparing the two structures, inactive and active, gives the picture of helix 6 swinging outward to admit the G protein.

✓ verbatim Kobilka, Nobel Lecture and biographical essay, Nobel Foundation (2012); Rasmussen et al., Nature 477:549 (2011).
Paul Hargrave

Paul Hargrave 1938–2025

His group sequenced bovine rhodopsin piece by piece and read seven membrane-spanning stretches out of the sequence, the plan that hormone receptors were later found to share.

Rhodopsin fills the membranes of rod cells, and like most membrane proteins it clumps in water. Hargrave's group, at Southern Illinois University and later the University of Florida, could not sequence it whole, so they cut it up. Cyanogen bromide breaks a protein chain at methionine and nowhere else, which turns one unmanageable protein into a fixed set of smaller pieces. Each piece went through Edman degradation, the chemistry that removes and identifies one amino acid at a time from the end of a chain. With S.-L. Fong and J. Hugh McDowell he published stretches of sequence through the early 1980s and the complete sequence of bovine rhodopsin in 1983. It held seven runs of water-avoiding residues, each long enough to cross a membrane, so the chain was drawn passing through the bilayer seven times with short loops on each face. When genes for hormone receptors were cloned a few years later, that drawing was what their sequences were compared with.

✓ verbatim Hargrave et al., Biophysics of Structure and Mechanism 9:235 (1983); Hargrave, the Friedenwald Lecture, Investigative Ophthalmology & Visual Science 42:3 (2001).
Yuri Ovchinnikov

Yuri Ovchinnikov 1934–1988

His Moscow institute sequenced bacteriorhodopsin and then bovine rhodopsin, reaching the seven-span arrangement of rhodopsin in the same years as Hargrave's group.

Ovchinnikov directed the Institute of Bioorganic Chemistry in Moscow from 1970. His laboratory's first success with a membrane protein was the sequence of bacteriorhodopsin, a light-driven proton pump from a salt-loving microbe, which crosses its membrane seven times and holds a molecule of retinal. The group then sequenced bovine rhodopsin with the tools Hargrave's group was also using and published in the Russian-language journal Bioorganicheskaya Khimiya. The 1983 paper with the finished sequence and its arrangement in the membrane is the third in a series titled Visual rhodopsin. In 1982 he had set the two proteins side by side in a review. They share the seven-pass plan and the retinal and little else: the microbial protein pumps protons, uses no G protein, and has no detectable sequence similarity to rhodopsin. That contrast became useful in 1986. The new β-adrenergic receptor sequence resembled the animal protein, the one that works through a G protein, and that pointed to G-protein coupling as what the family shares.

✓ verbatim Ovchinnikov, 'Rhodopsin and bacteriorhodopsin: structure-function relationships', FEBS Letters 148:179 (1982); Complete Dictionary of Scientific Biography, entry on Ovchinnikov.
Test yourself
Question 1 of 8 Score 0 / 0

How is the polypeptide chain of a G-protein-coupled receptor arranged in the plasma membrane?
Sources & Integrity

Every structure, slide, simulation, and clip on this page is listed below with the source it came from. None of the 3D structures is drawn by hand or by AI: each one loads by accession ID from RCSB PDB or PubChem. This panel is generated from the lesson's own data file, so what it lists is what the page loads.

Molecular structures live data

  • β2-adrenergic receptor bound to Gs — the active receptor holding its G protein — RCSB PDB 3SN6 · coordinates CC0 · Rasmussen et al., Nature 477:549 (2011)
  • β2-adrenergic receptor, inactive — the same receptor with an inverse agonist bound — RCSB PDB 2RH1 · coordinates CC0 · Cherezov et al., Science 318:1258 (2007); T4 lysozyme fusion
  • Bovine rhodopsin — the prototype of the seven-pass fold — RCSB PDB 1F88 · coordinates CC0 · Palczewski et al., Science 289:739 (2000)
  • Heterotrimeric G protein (Gi) with GDP — the resting switch — RCSB PDB 1GP2 · coordinates CC0 · Wall et al., Cell 83:1047 (1995)
  • Adrenaline (epinephrine) — one hormone, nine receptors — PubChem CID 5816 · PubChem CID 5816

Scientist profiles & quotes

  • Robert Lefkowitz — Lefkowitz, Nobel Lecture, 8 December 2012; Dixon, Kobilka et al., Nature 321:75 (1986).
  • Brian Kobilka — Kobilka, Nobel Lecture and biographical essay, Nobel Foundation (2012); Rasmussen et al., Nature 477:549 (2011).
  • Paul Hargrave — Hargrave et al., Biophysics of Structure and Mechanism 9:235 (1983); Hargrave, the Friedenwald Lecture, Investigative Ophthalmology & Visual Science 42:3 (2001).
  • Yuri Ovchinnikov — Ovchinnikov, 'Rhodopsin and bacteriorhodopsin: structure-function relationships', FEBS Letters 148:179 (1982); Complete Dictionary of Scientific Biography, entry on Ovchinnikov.

Quotes are verbatim from the cited source. Profile text is original to this site and was checked against the sources cited.

Animated explainer

The Receptor Decides the Answer · narration: an AI narrator voice (Gemini, Charon) reading narration written from this lesson's text · animation: AI-generated in code (Manim) by Claude, a schematic with no drawn molecular structures, plots computed from their equations · captions from the narration script.

Section films

  • Three Ways a Drug Moves a Receptor · narration: an AI narrator voice (Gemini, Charon) reading narration written from this lesson's text · animation AI-generated in code (Manim) by Claude, a schematic with no drawn molecular structures · captions from the narration script.

Images & figures

This lesson carries no raster figures. Every structure is rendered live from RCSB PDB or PubChem, and every plot is drawn from its equation in the browser.

integrity Machines check the mechanics. The build confirms that every local file exists, that every question's answer key points at a real choice, and that any AI-generated scientific figure names the person who checked it; a weekly job re-resolves every accession ID against its database. Questions are original; MCAT-style items are written for this site and are not real AAMC questions. Banned sources (the Tymoczko textbook, publisher decks, copyrighted figures) are not used.

Like I'm five

Your body sends messages, like adrenaline, through the blood. A message cannot get inside a cell, so the cell has a doorbell in its outer wall. When the message presses the doorbell, a helper just inside the wall wakes up and runs to tell the cell what to do. Different cells connect the same kind of doorbell to different helpers, so one message can speed up your heart and relax the muscles you breathe with.
High school

Many hormones and nerve signals cannot cross the cell membrane, so cells detect them with receptors that span it. The largest family of these receptors snakes through the membrane seven times. When a hormone such as adrenaline binds on the outside, the receptor changes shape on the inside and switches on a partner called a G protein. There are four main kinds of G protein, and each starts a different chain of events: one raises a messenger called cAMP, one lowers it, one raises calcium, and one changes the cell's shape. A cell's answer to a hormone depends on which receptor and which G protein it makes, which is how adrenaline can strengthen the heartbeat while relaxing the airways.
College

G-protein-coupled receptors are single chains with seven transmembrane helices, an extracellular amino end and a cytoplasmic carboxyl end. Agonist binding in the pocket among the helices stabilises an active conformation in which the cytoplasmic end of helix 6 swings outward and exposes a docking site for a heterotrimeric G protein. The receptor then acts as a guanine nucleotide exchange factor: Gα releases GDP, binds GTP, and separates from Gβγ. Four Gα families set the first step downstream. Gs stimulates adenylyl cyclase, Gi/o inhibits it, Gq/11 activates phospholipase C-β, and G12/13 activates exchange factors for Rho; free βγ has targets of its own, such as the GIRK potassium channel. Because tissues express different receptor subtypes coupled to different families, one ligand can produce opposite responses. Pertussis toxin ADP-ribosylates Gαi/o and uncouples it from receptors. Ligands are classed by the receptor conformation they favour: agonists the active one, inverse agonists the inactive one, and neutral antagonists neither.
Graduate

In the β2-adrenergic receptor–Gs complex the cytoplasmic end of TM6 is displaced outward by about 14 Å relative to the inactive, carazolol-bound receptor, and the carboxyl-terminal α5 helix of Gαs inserts into the cavity this opens. In the same structure the helical domain of Gαs has swung away from the Ras-like domain and exposed the nucleotide site, which fits the receptor's role as an exchange factor, because GDP release is the slow step of the unstimulated cycle. For purified Go, Higashijima and colleagues in Gilman's laboratory measured steady-state GTP hydrolysis at about 0.3 to 0.4 per minute against a catalytic step near 2 per minute, with GDP departure setting the pace. Pertussis toxin ADP-ribosylates a cysteine four residues from the carboxyl end of Gαi/o, in the segment that enters the receptor, which explains uncoupling with no change in GTPase chemistry. Ligand efficacy follows from a conformational equilibrium: a receptor with constitutive activity populates the active state without ligand, agonists and inverse agonists shift that population in opposite directions, and a neutral antagonist leaves it alone.

An open question is coupling selectivity. Roughly 800 human receptors choose among 16 Gα subunits, many choose more than one, and no simple sequence motif in the receptor predicts the choice. Structures of receptors bound to different G proteins differ in how far TM6 opens and how the α5 helix sits, but whether selectivity is set mainly by the shape of the final complex or by the kinetics of the steps that lead to it is unsettled.
Seven helices do not make a GPCR

Bacteriorhodopsin, from a salt-loving microbe, crosses its membrane seven times and holds a retinal that changes shape in light, as rhodopsin does. Ovchinnikov, whose institute sequenced both, compared them in a 1982 review. The likeness stops at the outline. Bacteriorhodopsin uses light to pump a proton across the membrane, works with no G protein, and has no detectable sequence similarity to rhodopsin. For a few years these were the only seven-pass proteins known, and Lefkowitz recalled in his Nobel lecture a speculation that seven spans marked light-sensitive molecules. The adrenaline receptor sequence of 1986 ended that idea.

— Ovchinnikov, FEBS Letters 148:179 (1982); Lefkowitz, Nobel Lecture, 8 December 2012.
An orphan gets a name

Tom Bonner's group at the National Institute of Mental Health searched rat brain DNA for sequences resembling known receptors and found a gene for a seven-pass protein. The sequence said nothing about what the protein detects. Such genes are called orphans. Its messenger RNA turned up in the brain regions and cell lines that carry cannabinoid binding sites, so cannabinoids were among the candidate ligands tried. Lisa Matsuda, first author of the 1990 paper, expressed it in cultured cells with no response to cannabinoids and looked for the properties Allyn Howlett's laboratory had described for cannabinoid action: lower cAMP with rising dose, activity in only one mirror-image form of the drug, and loss of the effect after pertussis toxin. The cloned receptor had them. It is now called CB1.

— Matsuda, Lolait, Brownstein, Young & Bonner, Nature 346:561 (1990).
The toxin that was named for the pancreas

Pertussis toxin was studied for years under another name. A protein purified from cultures of the whooping cough bacterium made rats release more insulin, so it was called islet-activating protein. Adrenaline normally restrains insulin release through α2 receptors, which couple to Gi, and the protein removed that brake. In 1982 Toshiaki Katada and Michio Ui reported that it works as an enzyme, moving ADP-ribose from NAD onto a membrane protein of about 41 kDa. That protein proved to be the α subunit of Gi. A toxin studied for its effect on the pancreas became a standard test for Gi/o coupling.

— Katada & Ui, Proc. Natl. Acad. Sci. USA 79:3129 (1982).

 ↑ Top