Developmental Neuroscience · Unit 8: Building synapses

Building the Neuromuscular Junction: Agrin, Lrp4 and MuSK

How does a motor axon and a muscle fibre build a synapse between them, and which side decides where it goes? This lesson follows the neuromuscular junction from the muscle's own band of receptor clusters, through the relay from nerve-derived agrin to Lrp4, MuSK, Dok7 and rapsyn, to the balance between agrin, which protects clusters, and acetylcholine, which disperses them. Afterwards you can predict what a knockout of each link does to the junction.

Watch first Agrin Protects What Muscle Starts · 2:49

A newborn mouse that lacks the muscle kinase MuSK cannot breathe. Its motor axons reach the diaphragm, but no neuromuscular junction forms.

The axon arrived. The muscle was there. Both cells were healthy, and still no synapse. So a synapse is not something that happens whenever an axon touches a target. It is built, by signals passing in both directions, and MuSK sits at the centre of that exchange. This lesson asks how the junction between a motor neuron and a muscle fibre is assembled, and which partner decides where it goes.

Chemical synapses: three signs of a synaptic specialization

Look at any chemical synapse in an electron micrograph and you will find three signs that the two membranes are specialized for each other:

  1. Small vesicles packed into the presynaptic terminal, holding transmitter.
  2. A narrow cleft between the two cells.
  3. A thickened postsynaptic membrane, where receptors and the proteins that hold them are concentrated.

The neuromuscular junction (NMJ), where a motor axon meets a skeletal muscle fibre, is where these were first worked out, and for good reasons. It is large, it lies outside the brain where you can reach it, and it is simple: one transmitter, acetylcholine, and one major receptor, the nicotinic acetylcholine receptor (AChR), a channel that opens when acetylcholine binds and lets cations into the muscle. (You met this receptor as a ligand-gated channel in that lesson.) The NMJ cleft also holds a sheet of basal lamina, the thin layer of extracellular matrix that wraps each muscle fibre. Keep that sheet in mind.

At a mature junction the receptors are packed under the nerve ending and sparse everywhere else on the fibre. Building the junction means solving that sorting problem: where do the receptors go, and who tells them?

The first contact is fast and physical. In culture, a growth cone from a frog spinal neuron adheres to a muscle cell rapidly after it touches; this is “the sticky synapse”. Adhesion holds the partners together. It does not by itself make a postsynaptic specialization.

The muscle was waiting

You might expect the nerve to choose where receptors gather. The first evidence against that came from culture: muscle cells grown with no neurons at all make small, spontaneous AChR clusters. When a motor axon is added, it does not seek out those clusters, but where it touches, far more clustering follows. And AChR activation is not the inducer: receptors still form synapses in mice that cannot make acetylcholine (Misgeld et al., PNAS 2005).

Culture leaves an objection. In an embryo, perhaps a nerve had marked the spot before anyone looked. In 2001 two groups answered this in the mouse. Xia Yang, Steven Burden and colleagues used mice whose motor neurons were eliminated genetically with diphtheria toxin, so that no motor axons reached the muscle. Weichun Lin, Joshua Sanes, Kuo-Fen Lee and colleagues also used mouse genetics. In both studies, the diaphragm still built a band of AChR clusters across its middle, and in the first study RNA for receptor subunits was also concentrated in that band. The muscle had laid down a prepattern.

The prepattern had limits. It needed MuSK: embryos lacking both motor axons and MuSK made receptors but no central band. It did not need agrin. And the nerve changed it: without innervation the band was wider. So the nerve does not need to choose the first position of the clusters, but contact makes far more clustering, narrows the band and keeps only the receptors beneath the terminal. Keep two statements apart. “AChR clusters can form without innervation” is true. “The nerve is not needed for a synapse” is false.

The agrin relay: from an empty sheath to a kinase

The search for the nerve’s signal began with a strange experiment. In 1978 Sanes, Lynn Marshall and U. J. McMahan damaged frog muscle so that the fibres and the nerve terminals degenerated, leaving only the sheaths of basal lamina. Regenerating axons returned to the old synaptic sites on the empty sheaths. In later work from the same laboratory, regenerating muscle fibres clustered receptors at the old sites even with the nerve kept out. The basal lamina remembered where the synapse had been.

Think of a building demolished down to its foundation, with a mark on the slab showing where the front door stood. The new builders read the mark. McMahan’s group went looking for the ink. They needed tissue denser in cholinergic synapses than muscle, and found it in the electric organ of the ray Torpedo. Fractions of electric organ were tested on cultured chick muscle cells, counting receptor clusters stained with fluorescent alpha-bungarotoxin, a snake toxin that binds the AChR. In 1987 they named the active protein agrin.

Motor neurons make agrin and send it into the basal lamina, but muscle makes agrin too. The difference is splicing. Neural agrin carries a short insert of a few amino acids at one splice site, and in culture assays on chick muscle cells that insert makes it more than a thousand times more potent at clustering receptors than the form without it, which muscle mainly makes. Mice lacking agrin had profoundly disrupted junctions.

Agrin had to be received. In 1996 the Regeneron group of David Glass and George Yancopoulos, with Burden, reported that mice lacking MuSK, a muscle-specific receptor tyrosine kinase, make no junctions, and that agrin acts through a complex containing MuSK. Like other receptor tyrosine kinases, MuSK signals when two kinase domains are brought together and phosphorylate each other (that lesson). Yet agrin did not bind MuSK directly. Something was missing.

The missing link was Lrp4, a large membrane protein of the LDL receptor family. Mice lacking Lrp4 were reported in 2006 to have NMJ defects resembling those of MuSK mutants. In 2008 two laboratories showed that Lrp4 is the agrin receptor. Natalie Kim and colleagues in Burden’s laboratory found that nonmuscle cells given Lrp4 bound neural agrin and that Lrp4 forms a complex with MuSK. Bin Zhang and colleagues in Lin Mei’s laboratory showed that purified neural agrin binds purified Lrp4, and that Lrp4 also binds MuSK.

Two more proteins finish the chain. Dok7, an adaptor inside the muscle, binds phosphorylated MuSK and boosts its activity; mice lacking Dok7 form no junctions. Rapsyn, a scaffold on the inner face of the membrane, holds the receptors together (that lesson). Mice lacking rapsyn make receptors but do not cluster them, though nuclei beneath the synapse still transcribe receptor genes.

neural agrin → Lrp4 → MuSK (with Dok7) → rapsyn → clustered AChR

Two forces: agrin protects, acetylcholine disperses

The prepattern raised a puzzle. If the muscle can cluster receptors with no agrin, why do mice lacking agrin lose most of their clusters? The answer is that the nerve brings a second signal. Acetylcholine, released at every terminal, disperses receptor clusters that agrin does not protect. The rule fits in one line: “Receptor clustering is a competition between two forces: facilitation by Agrin-MuSK and inhibition by ACh-AChRs.”

Activity also works through the nuclei. A muscle fibre is one long cell with many nuclei. When the nerve drives the fibre, receptor transcription is suppressed in nuclei away from the synapse, while the few nuclei beneath the terminal keep transcribing receptor genes, a pattern Burden’s group showed with transgenic mice in 1992. Cut the nerve and the suppression lifts: receptors appear along the whole fibre, and the denervated muscle becomes “supersensitive” to acetylcholine applied anywhere on it.

Design problem

You have three mouse lines: one lacking agrin, one lacking choline acetyltransferase (ChAT, the enzyme that makes acetylcholine, so its motor neurons release none), and the double knockout. Predict what AChR clusters look like at birth in each, and say what the double tells you about agrin’s job.

One way to do it

Use the two forces. Without agrin, MuSK still runs at its own lower level, but acetylcholine keeps dispersing clusters with nothing to protect them, so most clusters are lost. Without acetylcholine, nothing disperses clusters, so the muscle keeps more of them, spread over a wider band, and the nerve branches widely. In the double knockout, both forces are gone, and clusters return, built by the muscle’s own MuSK-dependent program.

The double is the telling line. If agrin’s job were to start clusters, removing acetylcholine could not bring them back. Clusters come back, so in the animal agrin’s main job is to protect clusters from dispersal by acetylcholine. Two mouse studies published in 2005, one from Sanes and Lichtman’s laboratories and one from Kuo-Fen Lee’s laboratory with Weichun Lin, reported this result.

NMJ vs. central synapses, and a junction that fails

The NMJ is a good guide to central synapses, but not a template. A central neuron receives thousands of inputs that use different transmitters, its synapses have no basal lamina in the cleft, and receptor clustering there uses other molecules, including neurexins, neuroligins and the scaffold PSD-95, taken up in that lesson. What carries over is the logic: each partner sends signals that make the other specialize, and stable synapses need both sides. How a muscle fibre that begins with several inputs ends with one is the subject of that lesson.

The NMJ also fails in a known way. In myasthenia gravis, the immune system makes antibodies against the junction, and muscles tire and weaken with use. Most patients have antibodies against the AChR. Some of the rest have antibodies against MuSK or Lrp4, two links from the chain above. Inherited mutations in several of the same genes, Dok7 among them, cause congenital myasthenic syndromes. The disease reads like a list of the knockouts.

How we measure it

Fluorescent alpha-bungarotoxin labeling

Alpha-bungarotoxin, a snake toxin, binds the muscle acetylcholine receptor tightly and nearly irreversibly. Coupled to a fluorescent dye, it shows where receptors sit. Labeling once in one colour and later in a second colour separates receptors that were already there from receptors added since.

Bioassay-guided purification

A tissue extract is split into fractions, and each fraction is tested on a living readout, here the number of receptor clusters on cultured muscle cells. Only fractions that keep the activity go forward. This is how agrin was found in Torpedo electric organ.

Knockout and motor-neuron-ablation mice

A knockout mouse lacks one gene in every cell. Mice that lack motor neurons, or whose motor axons never reach a muscle, show what the muscle can do alone. Comparing single and double knockouts places each gene in the pathway.

Whole-mount in situ hybridization

A labeled probe that pairs with one messenger RNA is applied to a whole piece of tissue, so you can see where the gene is transcribed. In embryonic diaphragm it showed that receptor-subunit RNA is concentrated in a central band.

Finish line

Test yourself

8 questions

Practice and MCAT-style, with explained answers.

Next · Lesson 36

Synapses in the Brain: Adhesion, Scaffolds and Astrocytes

A kidney cell that makes one neuronal protein can persuade an axon to build a nerve terminal on it.

Continue →

← Lesson 34 · Pruning Without Dying: Axons That Cut Themselves Back

Agrin bound to Lrp4 — the neural signal meets its muscle receptor
Agrin bound to Lrp4 — the neural signal meets its muscle receptor ● LIVE · RCSB PDB 3V64
open on RCSB PDB ↗
Loading live structure 3V64 from RCSB PDB…

Find the small agrin domain sitting against the ring-shaped propeller of Lrp4. The contact is where neural agrin is told apart from the muscle form.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
MuSK kinase domain — the enzyme that turns binding into a signal
MuSK kinase domain — the enzyme that turns binding into a signal ● LIVE · RCSB PDB 1LUF
open on RCSB PDB ↗
Loading live structure 1LUF from RCSB PDB…

A typical two-lobed tyrosine kinase. Look for the activation loop, which must be phosphorylated, and the segment near the membrane that holds the kinase off until then.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Torpedo acetylcholine receptor — the channel that the junction clusters
Torpedo acetylcholine receptor — the channel that the junction clusters ● LIVE · RCSB PDB 2BG9
open on RCSB PDB ↗
Loading live structure 2BG9 from RCSB PDB…

Five subunits ring a central pore that crosses the membrane. The binding sites for acetylcholine lie high in the part outside the cell, far from the gate.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Acetylcholine — the transmitter that opens the channel and disperses unprotected clusters
Acetylcholine — the transmitter that opens the channel and disperses unprotected clusters ● LIVE · PubChem CID 187
open on PubChem ↗
Loading live structure 187 from PubChem…

A small ester with a positively charged nitrogen. The charged end sits in a box of aromatic side chains in the receptor's binding site.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
U. J. McMahan

U. J. McMahan

Showed with Joshua Sanes and Lynn Marshall that the basal lamina of frog muscle keeps the address of the synapse after nerve and muscle are gone, then led the purification of a receptor-clustering activity from Torpedo electric organ that his group named agrin.

McMahan's question was where a synapse keeps its address. In a 1978 paper with Joshua Sanes and Lynn Marshall, frog muscle was damaged so that its fibres and nerve terminals degenerated, leaving the sheaths of basal lamina, the thin layer of extracellular matrix around each fibre. Regenerating axons returned to the old synaptic sites on those empty sheaths and began to differentiate there, with no muscle fibre beneath them. Something in the matrix marked the spot. A follow-up from the same laboratory found that regenerating muscle fibres gathered acetylcholine receptors at the old synaptic sites when the nerve was kept away. To find the molecule, the group needed tissue far richer in synapses than muscle. They used the electric organ of the ray Torpedo californica and followed an activity that clustered receptors on cultured chick muscle cells. The 1984 paper, with Earl Godfrey, Ralph Nitkin, Bruce Wallace and Lee Rubin, reported that activity, and a 1987 paper named the protein agrin. McMahan then proposed that motor neurons release agrin to organize the muscle membrane beneath them, a hypothesis that mouse genetics could test.

✓ verbatim Sanes, Marshall & McMahan, J. Cell Biol. 78:176 (1978); Godfrey et al., J. Cell Biol. 99:615 (1984); Nitkin et al., J. Cell Biol. 105:2471 (1987).
Joshua Sanes

Joshua Sanes

Co-led the basal lamina experiments, and his laboratory's mouse genetics showed that rapsyn and agrin are needed for the junction, that the first receptor clusters form without agrin or motor axons, and that agrin protects clusters from acetylcholine.

Sanes was first author of the 1978 basal lamina paper, and his later work turned the agrin hypothesis into tests in mice. In 1995, with John Merlie, Medha Gautam and colleagues, he reported that mice lacking rapsyn make acetylcholine receptors but fail to cluster them, while synapse-specific transcription of receptor genes continues. In 1996 the same collaboration reported that mice lacking agrin have profoundly disrupted neuromuscular junctions, with markedly fewer and smaller receptor clusters and abnormal nerve branching in the muscle. In 2001 a paper by Weichun Lin, Sanes, Kuo-Fen Lee and colleagues reported that the first postsynaptic differentiation in mouse muscle needs MuSK and rapsyn but not agrin or motor axons. Then in 2005, with Thomas Misgeld, Terrance Kummer and Jeff Lichtman, he used mice lacking both agrin and acetylcholine to argue that agrin's main job in the animal is to protect receptor clusters from a dispersing effect of the transmitter. Over those years agrin changed from the signal that builds the junction into the signal that lets the junction survive its own transmitter.

✓ verbatim Gautam et al., Cell 85:525 (1996); Lin et al., Nature 410:1057 (2001); Misgeld, Kummer, Lichtman & Sanes, Proc. Natl. Acad. Sci. USA 102:11088 (2005).
Steven Burden

Steven Burden

Showed that receptor genes are transcribed in the nuclei beneath the synapse, that mouse muscle forms a central band of receptor clusters with no motor neurons, and, with Natalie Kim, that Lrp4 is a receptor for agrin that forms a complex with MuSK.

Burden's work kept returning to the muscle side of the junction. In 1992, with Andrew Simon and Peter Hoppe, he used transgenic mice to show that expression of an acetylcholine receptor gene is restricted to the nuclei beneath the synapse. He was an author on the 1996 papers from David Glass, George Yancopoulos and colleagues showing that mice lacking MuSK form no junctions and that agrin acts through a complex containing MuSK. In 2001 his laboratory, with Xia Yang as first author, removed motor neurons from mouse embryos genetically and still found a central band of receptor clusters and receptor transcripts in the diaphragm. In 2008, with Natalie Kim and colleagues, it showed that nonmuscle cells given Lrp4 bind neural agrin, and that Lrp4 forms a complex with MuSK. Lin Mei's laboratory reached the same conclusion with purified proteins at almost the same time. Together the two papers filled the gap between agrin outside the muscle and the kinase that signals inside it.

✓ verbatim Yang et al., Neuron 30:399 (2001); Kim et al., Cell 135:334 (2008).
Test yourself
Question 1 of 8 Score 0 / 0

Which set lists the three signs of a synaptic specialization?
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

  • Agrin bound to Lrp4 — the neural signal meets its muscle receptor — RCSB PDB 3V64 · coordinates CC0 · Rat agrin fragment with the Lrp4 extracellular region, X-ray crystal structure (Zong et al., Genes & Development, 2012)
  • MuSK kinase domain — the enzyme that turns binding into a signal — RCSB PDB 1LUF · coordinates CC0 · MuSK cytoplasmic region, X-ray crystal structure (Till et al., Structure, 2002)
  • Torpedo acetylcholine receptor — the channel that the junction clusters — RCSB PDB 2BG9 · coordinates CC0 · Torpedo marmorata nicotinic acetylcholine receptor, electron microscopy of tubular crystals (Unwin, J. Mol. Biol., 2005)
  • Acetylcholine — the transmitter that opens the channel and disperses unprotected clusters — PubChem CID 187 · PubChem compound record for acetylcholine

Scientist profiles & quotes

  • U. J. McMahan — Sanes, Marshall & McMahan, J. Cell Biol. 78:176 (1978); Godfrey et al., J. Cell Biol. 99:615 (1984); Nitkin et al., J. Cell Biol. 105:2471 (1987).
  • Joshua Sanes — Gautam et al., Cell 85:525 (1996); Lin et al., Nature 410:1057 (2001); Misgeld, Kummer, Lichtman & Sanes, Proc. Natl. Acad. Sci. USA 102:11088 (2005).
  • Steven Burden — Yang et al., Neuron 30:399 (2001); Kim et al., Cell 135:334 (2008).

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

Animated explainer

Agrin Protects What Muscle Starts · 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.

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

A nerve tells a muscle when to squeeze by squirting a chemical at it. The muscle needs a big patch of catchers right under the nerve to feel it. The muscle starts making patches in its middle before the nerve even arrives. When the nerve comes, it hands over a protein called agrin that says "keep the patch here", and the patches anywhere else fall apart.
High school

At the junction between a motor nerve and a muscle, the nerve releases acetylcholine, and receptors on the muscle open to let ions in. Those receptors must be packed together under the nerve ending. In mouse embryos, the muscle forms a band of receptor clusters in its middle even when no nerve is present. When the nerve arrives, it releases a protein called agrin. Agrin binds a receptor called Lrp4, which switches on an enzyme called MuSK, and MuSK gathers a scaffold protein, rapsyn, that holds the receptors in place. Meanwhile, the nerve's acetylcholine breaks up clusters that agrin does not protect, so only the cluster under the nerve survives.
College

The neuromuscular junction uses one transmitter, acetylcholine, and one major receptor, the nicotinic acetylcholine receptor, a ligand-gated cation channel. Receptor clusters form without innervation, in culture and in a central band of embryonic mouse muscle that requires the receptor tyrosine kinase MuSK. Motor neurons secrete neural agrin, a splice form with a short insert that makes it far more active than muscle agrin. Neural agrin binds Lrp4, which associates with MuSK and stimulates its autophosphorylation; the adaptor Dok7 binds phosphorylated MuSK and strengthens its activity, and downstream signaling gathers rapsyn, which anchors receptors. Knockouts of agrin, Lrp4, MuSK, Dok7 or rapsyn each disrupt the junction. Acetylcholine acting on its receptors disperses clusters that agrin does not protect and, through activity, suppresses receptor transcription in nuclei away from the synapse, while subsynaptic nuclei keep transcribing.
Graduate

MuSK activation is the point where positive and negative inputs converge. The 2001 prepattern experiments placed MuSK, with rapsyn, upstream of nerve contact, and Kim and Burden later showed that raising MuSK expression shifts where clusters and synapses form and allows synapses to form in agrin-mutant mice. Agrin-Lrp4 binding thus acts on a kinase that already has some activity of its own, which explains why agrin is needed to keep and enlarge clusters more than to start them. The dispersing arm runs through receptor activation, and in mice it has been linked to the kinase Cdk5; the agrin and ChAT double mutant restores clusters, consistent with agrin acting mainly as an antagonist of this dispersal in vivo. Lrp4 also signals back to the nerve terminal in mouse work, so the same complex helps organize the presynaptic side.

An open question is what spatial cue centres the MuSK-dependent band in the first place. Candidates include intrinsic patterning of the muscle, Wnt signaling to MuSK and Lrp4, and the geometry of the fibres, and none is settled. A second open question is how synapse- specific transcription in subsynaptic nuclei is controlled, since removing neuronal neuregulin-1 in mice left the measured pattern in place.
A name earned in a dish

When McMahan's group reported the protein behind the clustering activity in 1987, they named it agrin, for its power to aggregate receptors. The name was earned in a dish. The purified material gathered acetylcholine receptors on cultured chick muscle cells, and it also gathered acetylcholinesterase and butyrylcholinesterase into the same patches. Antibodies against it labeled the basal lamina at synaptic sites in muscle, the place the frog regeneration experiments had pointed to nine years before.

— Nitkin, Smith, Magill, Fallon, Yao, Wallace & McMahan, J. Cell Biol. 105:2471 (1987).
A charged bead can do it

In culture, a latex bead coated with positive charge, laid on a frog muscle cell, can gather acetylcholine receptors beneath it. The result is a warning about the assay that found agrin. A receptor patch on a cultured cell shows that something reorganized the membrane; it does not show that the something is the signal the nerve uses. That is why later work asked whether agrin sits at the synapse and whether mice lacking it build normal junctions.

— Peng, Cheng & Luther, Nature 292:831 (1981).
Weak rabbits and a human disease

In 1973 Jim Patrick and Jon Lindstrom immunized rabbits with acetylcholine receptor purified from the electric organ of an eel, intending to raise antibodies. The rabbits became weak, and their weakness improved with a drug that blocks acetylcholinesterase, which is how myasthenia gravis responds in people. The experiment suggested that the human disease is an attack by antibodies on the receptor. Antibodies against the receptor were soon found in most patients.

— Patrick & Lindstrom, Science 180:871 (1973).

 ↑ Top