Cell Signaling · Unit 2: The parts — switches and wiring

Adaptors, Docking Proteins and Scaffolds

How can an activated receptor switch on Ras when it never touches Ras, and how can one kinase serve two pathways without mixing their messages? You will learn to tell adaptors, docking proteins and scaffolds apart, trace the Grb2 and Sos link from receptor to Ras, and say what a scaffold buys a cell.

Watch first Switching On Ras Without Touching It · 2:47

Ligand activates receptor… now what? A growth-factor receptor can switch on Ras without ever touching it, and the proteins that make that possible have no enzyme activity at all.

Ras is a small switch protein anchored to the inner face of the plasma membrane. When a growth factor binds its receptor, the receptor adds phosphates to tyrosines on its own tail. A phosphate on a tail is a message written where nothing can read it yet. The cell needs a coordinated message passed into the cytoplasm, and it gets one by recruiting proteins whose job is to put the right partners next to each other. This lesson sorts those proteins into three kinds, then asks what each kind buys the cell.

Zoomed out: the receptor recruits helpers

Each phosphotyrosine (a tyrosine carrying a phosphate) is a handhold for proteins with an SH2 domain, a small folded module that grips a phosphorylated tyrosine. Some of the proteins that arrive are enzymes. Others have no enzyme activity and only decide who stands next to whom. Three kinds of those are worth telling apart.

  • Adaptor. Links two partners and has no enzyme activity. Grb2 is the model.
  • Docking protein. A platform at the membrane that the receptor loads with many phosphotyrosines, each a landing site for a different partner. IRS-1 is the model.
  • Scaffold. Holds three or more members of one pathway on a single protein. Ste5 in yeast and KSR in mammals are the models.

The number of partners is a quick test: an adaptor typically bridges two, a scaffold holds several. A docking protein is the one that gets phosphorylated and so supplies the sites itself.

The adaptor: Grb2 and “son of sevenless”

In August 1992, Eve Lowenstein and colleagues in Joseph Schlessinger’s laboratory reported a protein with no catalytic machinery: one SH2 domain, two SH3 domains, nothing else. (An SH3 domain is another small module. It grips short proline-rich stretches.) They isolated its gene, found that it sticks by its SH2 domain to the activated receptors for epidermal growth factor and platelet-derived growth factor, and called it growth factor receptor-bound protein 2, or Grb2.

Here is how Grb2 does its job. Its SH2 domain clamps a phosphotyrosine on the receptor tail. Its two SH3 domains already hold a protein called Sos, with or without any phosphate. Sos has one task: it is an exchange factor that swaps the GDP on Ras for GTP, which flips Ras from off to on. The receptor holds Grb2, Grb2 holds Sos, and Sos now stands at the membrane where Ras is. Position does the rest.

receptor phosphotyrosine → Grb2 (SH2 on the receptor, SH3 on Sos) → Sos → Ras-GDP → Ras-GTP

In resting rat fibroblasts, injected Grb2 alone did nothing, and injected Ras protein alone did nothing. Injected together, they sent the cells into copying their DNA. That is biochemistry. The same chain turned up in two animals by an entirely different route, genetics.

In the compound eye of a fly, a receptor tyrosine kinase called sevenless decides whether one cell in each unit becomes R7, the seventh light-sensing cell. Michael Simon, in Gerald Rubin’s laboratory, ran a sensitized screen: weaken sevenless signaling first, then look for mutations that matter only against the weakened background. Out came alleles in a fly ras gene and in a gene resembling CDC25, the yeast protein that switches Ras on (Cell, November 1991). That gene is Sos, for son of sevenless.

In the nematode, Paul Sternberg’s group had placed let-23, a gene that looks like the EGF receptor, ahead of let-60, which turned out to be Ras. Stuart Clark, Michael Stern and Robert Horvitz cloned another worm gene on the same pathway, sem-5, in March 1992. Lowenstein’s abstract linked Grb2 to it.

So a fly screen, a worm clone and a rat-cell experiment, spread over nine months, landed on one chain. Grb2, the fly’s Drk and the worm’s Sem-5 are orthologs: one ancestral gene carried down three lineages. Ras, fly Ras1 and worm let-60 are orthologs too. The chain was sitting in animal cells before those lineages parted. What differs is what each animal does after Ras. In three of the six cells beside a larva’s gonad it builds a vulva. In a resting fibroblast it tells the cell to copy its DNA.

Docking proteins: one platform, many branches

Pathways branch through helper proteins like these, and the cleanest case is a docking protein. A receptor tail has only so much room. A docking protein solves that by being a second tail.

IRS-1 is the example. Insulin binds its receptor, and the receptor phosphorylates IRS-1. IRS-1 reaches the membrane with two modules of its own: a PH domain, which grips membrane lipids, and a PTB domain, which grips a phosphotyrosine motif on the activated receptor. Once it is phosphorylated at many tyrosines, it recruits PI 3-kinase, Grb2, the phosphatase Shp-2 and the adaptor Nck.

The few amino acids beside each phosphotyrosine decide which SH2 protein binds there. One loaded platform therefore sends the signal down several branches at once. Two of them:

Grb2 → Sos → Ras

PI 3-kinase → PDK1 → AKT

The first is the adaptor chain you just saw, hung from IRS-1 instead of from the receptor itself. The second ends in AKT, which regulates metabolism. Grb2 sits between two partners. IRS-1 is the thing that is phosphorylated, and it supplies the sites in return.

Scaffolds: what a cell gets from holding a pathway together

Watch The Yeast Protein That Holds Three Kinases · 1:47

Elaine Elion’s laboratory at Harvard Medical School was working out which parts of a yeast protein three enzymes stuck to. The protein was Ste5. Without it a yeast cell cannot mate. Yet the protein had no enzyme activity: it never put a phosphate on anything or took one off. In 1994 that came close to saying it had no mechanism.

Mating runs on three kinases in a row. Ste11 switches on Ste7, and Ste7 switches on Fus3, which halts the cell cycle so that the cell can fuse with a partner. The top one is shared, and the middle one is shared with the filament cascade. Ste11 also starts the cascade a yeast cell runs when the salt outside climbs too high, and the one it runs to grow in long filaments. One enzyme, three destinations, and a cell that smells pheromone mates and does nothing else.

Choi, Satterberg, Lyons and Elion reported in Cell in 1994 that Ste11, Ste7 and Fus3 bind different regions of Ste5, so one Ste5 molecule can hold all three at once. Two other laboratories reached the same result that year, both with the then-new yeast two-hybrid method: Michael Wigler’s group (PNAS, 2 August) and George Sprague’s (Genetics, November). A protein that catalyzes nothing earns its place by what it buys the cell.

Speed. Picture a pegboard with three tools hung within arm’s reach. A kinase and its target normally find each other by drifting at random through a crowded cytoplasm. Two proteins fastened to one scaffold are already touching.

Insulation. The Ste11 that a pheromone switches on is sitting against Fus3, nowhere near the kinase that would tell the cell it was drying out. A pathway-dedicated scaffold lets the same kinase serve different pathways, which Marcus and colleagues proposed in 1994 is how scaffolds inhibit cross-talk between kinase modules.

Location. Scaffolds carry their kinases to a place. InaD, built from PDZ domains (modules that grip the last few amino acids of a partner), gathers the light-response proteins of the fly photoreceptor in one spot. Paxillin collects Ras-pathway proteins at focal adhesions, the places where a cell grips its surroundings.

Scaffolds are often pictured as passive assembly platforms. In 2000, Levchenko, Bruck and Sternberg modeled a kinase cascade with a generic scaffold and showed that adding more scaffold does not keep raising the output. Past a certain amount, extra scaffold pulls the kinases onto separate scaffold molecules and the signal falls again. They proposed the name combinatorial inhibition for this effect. Then in 2006 Wendell Lim’s group showed that Ste5 is more than a tether: a Ste5 fragment bound to Fus3 drives Fus3 to put a phosphate on itself at one of its two activating sites.

Ste5 has no counterpart in humans. KSR, the scaffold of the Ras-MAPK pathway, and JIP-1 hold cascade kinases together in a comparable way but are not descended from it, so a search of the human genome for something that resembles STE5 would never have found them.

Place, and rewiring: scaffolds beyond the cascade

Here is a question about heart cells. A kinase reads a messenger, and an enzyme destroys that messenger. What holds the two close enough together for a local pool of messenger to exist?

The answer is a protein whose only job is to hold other proteins next to each other, an anchoring protein. In John Scott’s laboratory, Dodge and colleagues showed that mAKAP, the muscle-selective A-kinase anchoring protein, also holds a phosphodiesterase, PDE4, which destroys the messenger cyclic AMP. The kinase and its off-switch sit in the same complex, so the cyclic AMP that matters to this kinase is the cyclic AMP made right beside it, before it can move away. The complex also holds a target. Marx and colleagues reported that protein kinase A adds a phosphate to the cardiac ryanodine receptor, the channel that releases calcium for each heartbeat, and that the phosphate makes a protein called FKBP12.6 come off the channel. They also reported that this regulation is defective in failing hearts. Both findings are still debated, and some groups have not reproduced the loss of FKBP12.6.

Because a scaffold is a string of separate gripping surfaces, you can swap one surface for another that grips a different partner. Then a different input can reach a different output. You can hack scaffolds to mix inputs and outputs, and Lim’s laboratory has spent much of its time since building signaling proteins that no organism has had.

Design problem

Two pathways share one kinase, K. Input A must give output A only, and input B output B only. Design two different ways to keep the messages apart.

One way to do it

Design 1: separate scaffolds. Give each pathway its own scaffold. Scaffold A holds one copy of K, the activator that input A turns on, and the target that produces output A. Scaffold B holds another copy of K with B’s partners. The K that input A switches on is sitting against A’s target and nowhere near B’s. This is the Ste5 logic. Its weak point is dose: too much scaffold splits every complex into pieces, so the amount has to be tuned.

Design 2: separate places. Anchor K at two sites with two anchoring proteins. Input A makes the messenger at site A and input B makes it at site B. At each site K sits beside its own target and beside an enzyme that destroys the messenger, so the messenger from the other site is cleared before this kinase meets it. This is the mAKAP logic. Its weak point is the anchor: lose the enzyme from a complex and local control goes with it.

Other answers work. A good one says what keeps the messages apart and what breaks when a part goes missing.

How we measure it

Sensitized genetic screen

You weaken one pathway so that it is only just working, then look for mutations that matter only against that weak background. A mutation that does nothing in a normal animal can now tip the pathway over. Simon and Rubin used it in the fly eye and found ras and Sos.

Binding-based cloning and microinjection

You find a gene by what its protein sticks to, for example a phosphorylated growth-factor receptor, then inject the purified protein into resting cells and score whether they copy their DNA. It tells you that a protein binds a receptor and that it feeds a pathway that drives growth.

Co-purification of protein complexes

You pull one protein out of a cell extract and see what comes with it. Elion's group used it to show that Ste5 copurifies with Ste11, Ste7 and Fus3.

Yeast two-hybrid assay

You fuse two proteins to the two halves of a transcription factor in yeast. If the proteins bind each other, the halves come together and switch on a reporter gene. Printen and Sprague used it to show that Ste5 interacts with each kinase of the mating cascade.

Finish line

Test yourself

8 questions

Practice and MCAT-style, with explained answers.

Next · Lesson 11

Lipids as Anchors and Addresses

How does a protein with no membrane-spanning segment end up stuck to a membrane, and how does a lipid label tell it where?

Continue →

← Lesson 9 · Protein Modules: SH2, SH3, PTB, PDZ and PH

Grb2, full-length adaptor — SH2 and SH3 domains in one chain
Grb2, full-length adaptor — SH2 and SH3 domains in one chain ● LIVE · RCSB PDB 1GRI
open on RCSB PDB ↗
Loading live structure 1GRI from RCSB PDB…

Find the one SH2 domain flanked by two SH3 domains. There is no catalytic site anywhere in the molecule: every surface is a place to hold something.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Sos bound to Ras — the switch-thrower on its target
Sos bound to Ras — the switch-thrower on its target ● LIVE · RCSB PDB 1BKD
open on RCSB PDB ↗
Loading live structure 1BKD from RCSB PDB…

Sos is the protein that Grb2 holds beside the membrane. Look at how it pries open the nucleotide-binding pocket of Ras.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
PSD-95 PDZ domain with bound peptide — a gripping surface of a scaffold
PSD-95 PDZ domain with bound peptide — a gripping surface of a scaffold ● LIVE · RCSB PDB 1BE9
open on RCSB PDB ↗
Loading live structure 1BE9 from RCSB PDB…

A PDZ domain grips the last few amino acids of a partner protein. Scaffolds such as InaD are built from several of these.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Joseph Schlessinger

Joseph Schlessinger born 1945

Led the laboratory that isolated Grb2, a protein built only from SH2 and SH3 domains, and showed that it links activated growth-factor receptors to Ras.

In August 1992 a paper from Joseph Schlessinger's laboratory at NYU Medical Center, with Eve Lowenstein as first author, described a protein that had no catalytic machinery at all: one SH2 domain, two SH3 domains, nothing else. The team isolated its gene and found that it sticks, by its SH2 domain, to the activated receptors for epidermal growth factor and platelet-derived growth factor, and named it growth factor receptor-bound protein 2, or Grb2. Injected alone into resting rat fibroblasts, Grb2 did nothing, and Ras protein alone did nothing either. Injected together, they sent the cells into copying their DNA. The abstract then pointed at a worm. Grb2 matched sem-5, a nematode protein cloned that March, and sem-5 acted on the same pathway as the worm receptor and the worm Ras gene. A fly screen had already found Sos the year before. Knowing the order of the parts told drug hunters where a receptor's signal could be cut. Schlessinger co-founded the biotechnology companies Sugen (1991), Plexxikon (2001) and Kolltan.

✓ verbatim Lowenstein et al., Cell 70:431 (1992), the Grb2 paper; the Cell abstract names the worm homologue sem-5.
Michael Simon

Michael Simon active 1991 onward

In Gerald Rubin's laboratory, ran a sensitized genetic screen in the fly eye that found fly ras and Sos, and showed that Sos encodes a relative of CDC25, the yeast activator of Ras.

In the compound eye of a fruit fly, a receptor tyrosine kinase called sevenless decides whether one cell in each unit becomes R7, the seventh light-sensing cell. At Berkeley, Michael Simon worked on that pathway in Gerald Rubin's laboratory. A screen on healthy flies would miss the many mutations that matter only when the pathway is already under strain. So Simon and Rubin first weakened sevenless signaling, then looked for mutations that mattered only against the weakened background. The screen returned alleles in a fly ras gene and in a gene resembling CDC25, the yeast protein that switches Ras on. The paper appeared in Cell in November 1991. That second gene was Sos, for son of sevenless, which Utpal Banerjee's laboratory had first isolated, as a dominant mutation, earlier in 1991. Simon's paper showed that it encodes a relative of CDC25. It handed the receptor-to-Ras chain a likely missing link before the mammalian adaptor had been found.

✓ verbatim Simon et al., Cell 67:701 (1991), the screen for dominant enhancers of a temperature-sensitive sevenless allele.
Elaine Elion

Elaine Elion active 1984 onward

Showed that Ste5, a yeast protein with no enzyme activity, holds all three kinases of the mating cascade at once.

Elaine Elion's laboratory at Harvard Medical School was working out which parts of a yeast protein three enzymes stuck to. The protein was Ste5. Without it a yeast cell cannot mate, yet the protein never put a phosphate on anything or took one off. In 1994 that came close to saying it had no mechanism. Two-hybrid mapping showed that Ste11, Ste7 and Fus3 bind different regions of Ste5, so one Ste5 molecule can hold all three. That explained how a cell can run three cascades from one shared top kinase and still answer a pheromone with mating alone. Two other laboratories reached the same result in the same year. Elion kept working on Ste5 for nearly three decades, and the question it opened changed from which enzyme a protein carries to what it holds.

✓ verbatim Choi, Satterberg, Lyons and Elion, Cell 78:499 (1994); Elion, Grisafi and Fink, Cell 60:649 (1990) on FUS3.
Wendell Lim

Wendell Lim active 1990s onward

Showed that Ste5 does more than tether: binding it changes the shape of Fus3 and helps the kinase activate itself. His laboratory builds signaling proteins no organism has had.

By the time Wendell Lim's laboratory at the University of California, San Francisco went back to Ste5, the passive tether was the usual picture of a scaffold. That was the obstacle. Roby Bhattacharyya, working with Attila Reményi, Matthew Good, Caleb Bashor and Arnold Falick, reported in 2006 that a fragment of the scaffold, bound to Fus3, drives it to put a phosphate on itself at one of its two activating sites. Showing that took structural and biochemical work, because the obvious explanation was already on the table: hold two enzymes close and the reaction speeds up. The team solved a crystal structure of a Ste5 fragment bound to Fus3 and measured Fus3's rate of self-phosphorylation, which rose about fifty-fold. Lim had come to signaling sideways, from biophysical chemistry and protein structure, and his laboratory has since spent much of its time building signaling proteins that no organism has had. A scaffold, on that view, is a set of parts that can be recombined.

✓ verbatim Bhattacharyya et al., Science 311:822 (2006), on Ste5 and Fus3; Bhattacharyya, Reményi, Yeh and Lim, Annual Review of Biochemistry (2006), on modular domains.
Test yourself
Question 1 of 8 Score 0 / 0

Which description fits an adaptor such as Grb2?
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

  • Grb2, full-length adaptor — SH2 and SH3 domains in one chain — RCSB PDB 1GRI · coordinates CC0 · Crystal structure of the mammalian Grb2 adaptor (Maignan et al., Science 1995).
  • Sos bound to Ras — the switch-thrower on its target — RCSB PDB 1BKD · coordinates CC0 · Ras in complex with the catalytic region of human Sos (Boriack-Sjodin et al., Nature 1998).
  • PSD-95 PDZ domain with bound peptide — a gripping surface of a scaffold — RCSB PDB 1BE9 · coordinates CC0 · Third PDZ domain of PSD-95 bound to a peptide tail (Doyle et al., Cell 1996).

Scientist profiles & quotes

  • Joseph Schlessinger — Lowenstein et al., Cell 70:431 (1992), the Grb2 paper; the Cell abstract names the worm homologue sem-5.
  • Michael Simon — Simon et al., Cell 67:701 (1991), the screen for dominant enhancers of a temperature-sensitive sevenless allele.
  • Elaine Elion — Choi, Satterberg, Lyons and Elion, Cell 78:499 (1994); Elion, Grisafi and Fink, Cell 60:649 (1990) on FUS3.
  • Wendell Lim — Bhattacharyya et al., Science 311:822 (2006), on Ste5 and Fus3; Bhattacharyya, Reményi, Yeh and Lim, Annual Review of Biochemistry (2006), on modular domains.

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

Animated explainer

Switching On Ras Without Touching It · 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

  • The Yeast Protein That Holds Three Kinases · 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

Imagine you want to flip a light switch across the room, but your arms are too short. A friend holds your hand with one hand and holds a long stick with the other, and the stick reaches the switch. Cells do this too. Small helper proteins do no work of their own. They just hold other proteins together so a message can pass from one to the next, and some of them hold a whole team in one spot so the team can work fast.
High school

When a hormone or growth factor lands on a receptor, the receptor adds phosphate tags to its own tail. The cell reads those tags with helper proteins that have no enzyme activity. An adaptor connects two proteins, a docking protein offers many tags for many partners, and a scaffold holds several steps of one pathway on a single frame. Holding things together makes the steps faster, keeps two pathways that share a part from getting mixed up, and puts the pathway in the right place in the cell.
College

Activated receptor tyrosine kinases autophosphorylate, and the phosphotyrosines recruit SH2-domain proteins. The adaptor Grb2 has one SH2 domain that binds the receptor and two SH3 domains that constitutively bind the Ras exchange factor Sos, so recruiting Grb2 brings Sos to the membrane where Ras sits. Docking proteins such as IRS-1 are themselves phosphorylated on many tyrosines and branch the signal to several effectors, including PI 3-kinase and Grb2. Scaffolds such as Ste5 bind several kinases of one cascade at once. They raise the local concentration of enzyme and substrate, they keep a kinase shared by several pathways, such as Ste11, from crossing over between them, and they can anchor a pathway at a site such as a focal adhesion or beside an enzyme that removes a messenger, as AKAPs do with PDE4.
Graduate

A scaffold with n binding sites forms a productive complex only when all n partners occupy one molecule. If the scaffold is in large excess over each partner, the partners distribute onto different scaffold molecules, and the fraction of fully loaded scaffolds falls toward zero. Output is therefore biphasic in scaffold concentration, which Levchenko and colleagues called combinatorial inhibition, and scaffold abundance becomes a tuning parameter rather than a gain control. Scaffolds are not only tethers. Ste5 binding to Fus3 changes the conformation of Fus3 and promotes its autophosphorylation, so some of the rate enhancement comes from allostery and not from proximity alone, and separating the two takes structural and biochemical dissection. Scaffolds with the same function, such as KSR and JIP-1, are not descended from Ste5. One open question is how much of the benefit of other scaffolds is proximity and how much is allostery.
The worm gene that turned out to be Ras

In 1990, Min Han, a postdoctoral fellow in Paul Sternberg's laboratory at Caltech, cloned the worm gene let-60, which controls whether cells become vulval tissue. Genetics had placed let-60 downstream of let-23, a worm gene that looked like the human receptor for epidermal growth factor. Over its first 164 amino acids the let-60 protein was 84 percent identical to Ras proteins from other species. Robert Horvitz's laboratory at MIT reported the same finding within a week. The worm's receptor chain ended in Ras.

— Han and Sternberg, Cell 63:921 (1990); Han, Aroian and Sternberg, Genetics 126:899 (1990); Aroian et al., Nature 348:693 (1990); Beitel, Clark and Horvitz, Nature 348:503 (1990).
A protein with nothing to write down

Knock out STE5 and a yeast cell cannot mate. Ask what the Ste5 protein does and there was nothing to write down: it never puts a phosphate on anything and never takes one off. Its biochemical function was unknown. In 1994 three laboratories showed within about three months that it binds the mating kinases. Wigler's group and Sprague's group used the yeast two-hybrid method, and Elion's group combined two-hybrid mapping with co-purification.

— Choi et al., Cell 78:499 (1994); Marcus et al., PNAS 91:7762 (1994); Printen and Sprague, Genetics 138:609 (1994).
More scaffold, less signal

You might expect that more scaffold always means more signal. In 2000, Levchenko, Bruck and Sternberg modeled a kinase cascade with a generic scaffold and found otherwise. Past a certain concentration, extra scaffold pulls the kinases apart onto separate scaffold molecules, one kinase to a scaffold, and the output falls again. The effect had been described for protein complexes in general by Bray and Lay in 1997, and Levchenko and colleagues proposed the name combinatorial inhibition. Levchenko and colleagues noted that this could settle a conflict in the literature, because the scaffold JIP-1 had been reported both to inhibit and to support signaling.

— Levchenko, Bruck and Sternberg, PNAS 97:5818 (2000).

 ↑ Top