Developmental Neuroscience · Unit 6: Axons find their way

Pioneers, Guideposts and the Logic of Guidance Cues

How does a growth cone know which way to go? This lesson sorts guidance cues into four classes, then shows how a long route is cut into short legs by pioneer axons, guidepost cells and the bundles that later axons follow, using grasshopper, fly and chick experiments. Afterwards you can read a loss-of-function and a gain-of-function phenotype together, and explain why one cue can attract one axon and repel another.

Watch first A Long Route in Short Legs · 2:38

In the growing leg of a grasshopper embryo, the first two axons to head for the central nervous system have no other axons to follow. Kill a particular guidepost cell along their route before they get there, and they lose their way.

That result, from David Bentley and Michael Caudy in 1983, is a clue to how a growth cone navigates. It does not carry a map. It reads its surroundings one stretch at a time, and some of what it reads is other cells. This lesson asks what those signals are, how a long route gets broken into short pieces, and why the same signal can mean different things to different axons.

What provides directional information to growth cones?

The growth cone is the moving tip of a growing axon: a fan of actin with fingers called filopodia, backed by microtubules from the axon shaft. It advances when adhesion grips the substrate, and it turns when that balance differs between its two sides (see that lesson). The question here is what tells it which side to favour.

The answers sort into four classes, set by two questions. Is the cue fixed to a cell or to the matrix between cells, so that it acts on contact, or is it released and spread through the tissue, so that it acts at a distance? And does it draw the growth cone in or push it away?

  • Contact attraction. Laminin in the extracellular matrix; adhesion molecules such as L1 on other axons.
  • Contact repulsion. Ephrins on cell surfaces, which make growth cones carrying the matching Eph receptor collapse and retreat.
  • Long-range attraction. Netrin-1, secreted by the floor plate at the ventral midline of the spinal cord, which draws commissural axons toward it.
  • Long-range repulsion. Slit at the same midline, and secreted semaphorins such as Sema3A.

The classes describe how a cue behaves in a given experiment, not a fixed property of the molecule. Netrin can act at short range when it is held on cell surfaces, and later lessons take each family in turn.

The cleanest test of a long-range cue is a pipette. In 1979 Gundersen and Barrett set a fine pipette releasing nerve growth factor (NGF) beside growing axons of chick sensory neurons, and the growth cones turned toward it within about twenty minutes. The turn was more than faster growth, which is what makes it chemotaxis: steering up a gradient.

The genome holds far too few genes to name every connection in a brain (the arithmetic is in that lesson). The genome gets around this in two ways. Cues act in combination, so a few families give many distinct readings. And the route is shortened.

Break the long journey into manageable legs

A long trajectory is hard to steer in one go. A trail with signposts every few hundred metres is easy. Axons use the same trick. The neuron’s own ability to grow is not enough: growing axons “need signs.”

The first axons through a territory are pioneers. They travel when the tissue is small and distances are short, and they have nothing to follow. Later axons, the followers, often grow along them. Along the pioneers’ route sit guidepost cells and intermediate targets: places where the growth cone stops, reads something and changes course.

The worked case is the grasshopper leg. Near the tip of the developing leg, a pair of sensory neurons called Ti1 sends axons toward the central nervous system. The route has several turns, and at some of them sit young neurons that the Ti1 growth cones touch just before they turn. One of these is Cx1.

Bentley and Caudy wanted to know whether the growth cones need Cx1. Insect neurons carry a surface sugar that antibodies against horseradish peroxidase recognize, so antibodies of this kind can mark identified neurons. Bentley and Caudy destroyed Cx1 with focused light before the Ti1 growth cones arrived, then cultured the embryos and stained the axons. The pioneers failed to take the normal route and sent branches in abnormal directions.

The conclusion is narrow, and that is its strength. At this turn, in this preparation, Cx1 is needed. The experiment does not say what Cx1 provides, whether adhesion or a signal, and later work by Caudy and Bentley found that the full Ti1 route is built from several legs, read partly from guideposts and partly from boundaries in the surrounding tissue. A guidepost is one important cue among several.

The labeled pathway hypothesis

Followers do not grow along just any axon. When axons grow bundled together they are fasciculated; when one leaves its bundle it defasciculates.

Jonathan Raper, Michael Bastiani and Corey Goodman showed how selective this is, again in grasshopper embryos, whose central neurons can be named and filled with dye one at a time. A neuron called G grows into a region crossed by many longitudinal axon bundles, joins one of them, the A/P fascicle, and grows forward along the P axons within it. Its filopodia reach other bundles, yet it chooses this one. When the team killed the P neurons before G arrived, G failed to extend far along its normal route and did not adopt any other bundle within reach. This led to the labeled pathway hypothesis: each bundle carries surface labels, and each growth cone has receptors for some of them. The ablation alone cannot tell whether P axons instructed G or simply offered the one usable surface.

The labels were found in the fruit fly, Drosophila. Fasciclin II is a member of the immunoglobulin (Ig) superfamily of adhesion molecules and binds Fasciclin II on a neighbouring axon. Its vertebrate relative is NCAM. The two phenotypes are complementary. Without Fasciclin II, axons that normally travel together in certain pathways drift apart. With extra Fasciclin II in motor neurons, motor axons that should leave their nerve at a branch point stay stuck in it and bypass their muscles. Opposite manipulations give opposite results, so the amount of adhesion between axons is what matters.

A second fly gene, beaten path (beat), encodes a protein secreted by motor neurons. In beat mutants, motor axons fail to leave their nerve at choice points.

Design problem

A Drosophila motor axon in a beat mutant fails to leave its nerve. Predict the phenotype of a beat; fasII double mutant, and say what it tells you about how Beat works.

One way to do it

Consider two models. In the first, Beat steers the axon toward its muscle. Then removing Fasciclin II should not help: the axon would still have no instruction to go to the muscle, and the double mutant would look like beat or worse. In the second, Beat lets an axon leave by weakening the Fasciclin II adhesion that holds it to its neighbours. Then removing Fasciclin II should restore defasciculation, and the double mutant would look closer to normal.

The second is what Fambrough and Goodman reported in 1996: reducing Fasciclin II suppressed the beat phenotype. So Beat acts by opposing adhesion between axons, not by pointing the way. A further check follows: extra Fasciclin II in a normal fly should mimic a beat mutant, and it does: extra Fasciclin II on motor neurons produced bypass phenotypes (Lin and Goodman, 1994).

Vertebrates show the same logic at choice points. Kathryn Tosney and Lynn Landmesser labeled chick motor axons growing into the hindlimb. At the plexus, where motor axons sort into dorsal and ventral nerve trunks, and where muscle nerves branch off, the growth cones were larger and more complex and turned more often. In straight stretches they were small and simple. Later time-lapse imaging, in mouse retinal axons, matched complex shapes to growth cones that had paused. And growth cone pausing is closely related to axon branching. Branches tend to form where a growth cone has stopped to read its surroundings.

Do growth cones like more adhesive or less adhesive substrate?

Given a choice in culture, growth cones tend to follow the more adhesive surface (Paul Letourneau, 1975, with chick sensory neurons). That makes it tempting to say adhesion is what they want. But do growth cones grow faster on the more adhesive surface? Often not: neurons can grow more slowly on more adhesive substrates, and in several culture studies how sticky a surface is turned out to be a poor predictor of growth rate.

The clutch model explains why. With too little grip, the actin network slips backward and the edge does not advance. With too much, the growth cone cannot let go of old adhesions to move on. Growth is fastest at intermediate adhesion, and the surface also signals.

Laminin is the clearest case. It is a large glycoprotein of basement membranes, and in the early 1980s several laboratories found that it strongly promotes neurite growth in culture. Neurons read laminin through integrins, receptors made of an alpha and a beta chain. In 1986 Kevin Tomaselli, Louis Reichardt and John Bixby blocked integrins with antibodies on dishes where cell bodies could still attach through a second coating. Outgrowth on laminin stopped while the cells stayed down, so integrins were needed for extension, not only to hold the cell on. Integrins link the matrix to actin and signal through Rho-family GTPases (that lesson).

Growth cones carry three main families of adhesion molecule:

  • Cadherins, such as N-cadherin, which need calcium and bind the same cadherin on another cell.
  • Ig-superfamily CAMs, such as NCAM and L1, which bind the same molecule or a different one. Fasciclin II and neuroglian are their fly relatives.
  • Integrins, which bind matrix proteins such as laminin. Mammals have 18 alpha and 8 beta chains, forming 24 known pairs.

A uniform coat of laminin permits growth but gives no direction. A border or a gradient can.

Attraction and repulsion can flip.

The same cue can attract one axon and repel another. Netrin-1 attracts growth cones that carry the receptor DCC alone and repels those that carry DCC with a second receptor, UNC5. Inside a single growth cone, the answer can change too. In cultured Xenopus spinal neurons, lowering cAMP signaling turned attraction to netrin-1 into repulsion. Xenopus retinal growth cones are attracted by netrin-1 on their own, but laminin turns that response into repulsion.

So a cue is a message, and the receptors and internal state of the growth cone decide what it means. That is why a few cue families can wire many different axons: a cell that changes its receptors or its cAMP level reads the same tissue differently.

There is one more reason to expect several cues at each choice point. An axon cannot easily take back what it has built. At choice points, a contact cue usually works beside a long-range one, and an attractant beside a repellent. If one system fails, the growth cone is still not sent somewhere it cannot recover from. The next lessons take these systems apart: netrin and DCC, Slit and Robo at the midline and semaphorins. For the receptor families themselves, see that lesson.

How we measure it

Killing an identified cell before an axon arrives

In insect embryos individual cells can be recognized and labeled, then destroyed with focused light or a microelectrode before a growth cone reaches them. Comparing the route with an operated control on a nearby spot asks whether that one cell is needed for a normal turn.

Dye filling of identified neurons

A neuron is injected with a fluorescent dye such as Lucifer Yellow or with horseradish peroxidase, which fills its axon and growth cone. Because the same named neuron can be found in every embryo, you can ask which bundle it chooses and whom its growth cone touches.

Loss-of-function, gain-of-function and double mutants in Drosophila

Remove a gene, express it in extra cells or at higher levels, or combine two mutations, then stain the motor axons with an antibody. Opposite phenotypes from loss and gain point to a quantity the gene sets; a double mutant can show that one gene works by opposing another.

Substrate choice and pipette turning assays

Neurons are grown on dishes coated with patterns of different substrates, or near a pipette releasing a cue. Which surface the growth cones follow, how fast they grow, and whether they turn toward or away from a source each test a different property of the cue.

Finish line

Test yourself

8 questions

Practice and MCAT-style, with explained answers.

Next · Lesson 23

Netrin and DCC: One Cue, Two Answers

The same protein pulls one axon toward the midline and pushes another away. What decides?

Continue →

← Lesson 21 · One Axon, Many Dendrites: Neuronal Polarity

Laminin alpha 2 LG4-5 domain pair — a part of the basement membrane protein that supports axon growth
Laminin alpha 2 LG4-5 domain pair — a part of the basement membrane protein that supports axon growth ● LIVE · RCSB PDB 1DYK
open on RCSB PDB ↗
Loading live structure 1DYK from RCSB PDB…

Two globular domains from the end of one laminin chain. Domains like these bind cell surface receptors; the integrin-binding site for neurons sits in a nearby region of the full cross-shaped molecule, which is far larger than this fragment.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
N-cadherin, two-domain fragment — a calcium-dependent adhesion molecule on growth cones
N-cadherin, two-domain fragment — a calcium-dependent adhesion molecule on growth cones ● LIVE · RCSB PDB 1NCJ
open on RCSB PDB ↗
Loading live structure 1NCJ from RCSB PDB…

Find the calcium ions at the joint between the two domains. They hold the domains rigid; take the calcium away and cadherins become floppy and stop binding the same cadherin on another cell.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
NCAM Ig1-3 fragment — a vertebrate relative of fly Fasciclin II
NCAM Ig1-3 fragment — a vertebrate relative of fly Fasciclin II ● LIVE · RCSB PDB 1QZ1
open on RCSB PDB ↗
Loading live structure 1QZ1 from RCSB PDB…

Three immunoglobulin-like domains in a row. Like Fasciclin II, NCAM can bind the same molecule on a neighbouring membrane, which is how such proteins hold axons together in a bundle.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Integrin alphaVbeta3 extracellular segment — the heterodimer shape of matrix receptors
Integrin alphaVbeta3 extracellular segment — the heterodimer shape of matrix receptors ● LIVE · RCSB PDB 1JV2
open on RCSB PDB ↗
Loading live structure 1JV2 from RCSB PDB…

Two chains, alpha and beta, fold back on themselves in this bent form. The ligand-binding head is formed by both chains together. Neurons use other pairs, such as alpha3beta1 and alpha6beta1, to bind laminin.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
David Bentley

David Bentley

Worked out the route of the Ti1 pioneer axons in the grasshopper embryo's leg and, with Michael Caudy, showed in 1983 that killing a guidepost cell at one turn before the pioneers arrived disrupted their normal route.

Bentley worked on grasshopper embryos, where a developing leg holds only a few nerve cells and each can be recognized from one embryo to the next. Near the tip of the leg, a pair of sensory neurons called Ti1 sends the first axons toward the central nervous system. Nothing has gone that way before them, so they are pioneers, and their route has a series of turns. At one turn sit young neurons called Cx1, which the Ti1 growth cones normally touch before they change direction. With Michael Caudy he made Cx1 visible and destroyed it with focused light before the pioneers reached it, then cultured the embryos. The Ti1 growth cones failed to take the normal route and branched in abnormal directions. Later work with Caudy showed that the whole route is built from several legs, each read from different cues. The experiment made a guidepost something you could remove and test.

✓ verbatim Bentley & Caudy, Nature 304:62 (1983); Caudy & Bentley, J. Neurosci. 6:1781 (1986).
Corey Goodman

Corey Goodman

With Jonathan Raper and Michael Bastiani, showed that an identified grasshopper growth cone chooses one specific axon bundle among many, which framed the labeled pathway hypothesis; his laboratory then used Drosophila genetics to identify adhesion molecules such as Fasciclin II and the secreted protein Beat that control when axons bundle and unbundle.

Goodman's early work also used grasshopper embryos, whose central nervous system is built from neurons that can be named and filled with dye one at a time. With Jonathan Raper and Michael Bastiani he followed a neuron called G. Its growth cone reaches a region crossed by many longitudinal axon bundles, yet it joins one, the A/P fascicle, and grows along the P axons within it. Electron microscopy showed G's growth cone touching P axons more than their neighbours. When the team killed the P neurons before G arrived, G failed to extend far along its normal route and did not adopt other bundles within reach. Such results led to the labeled pathway hypothesis: bundles carry surface labels that growth cones recognize. His laboratory moved to Drosophila to find the labels genetically, and showed that Fasciclin II holds some axons together while the secreted protein Beat lets motor axons leave their nerve. The fly work turned a cellular idea into genes that could be removed and added back.

✓ verbatim Raper, Bastiani & Goodman, J. Neurosci. 3:31 (1983) and 4:2329 (1984); Lin & Goodman, Neuron 13:507 (1994); Fambrough & Goodman, Cell 87:1049 (1996).
Lynn Landmesser

Lynn Landmesser

With Kathryn Tosney, showed that chick motor growth cones change shape and turn where routes divide in the developing hindlimb, so that choice points are places of active decision rather than passive channeling.

Landmesser's laboratory studied how motor axons from the chick spinal cord reach the right muscles in the hindlimb. By the 1980s her group had shown that these early projections are already precise; the right muscle was not reached mainly by making errors and removing them later. With Kathryn Tosney she asked what single growth cones do on the way. They injected horseradish peroxidase, an enzyme that can be stained, into the spinal cord to label motor axons, then cut the embryos into serial sections and drew the tips one by one. At the plexus, where motor axons sort into dorsal and ventral nerve trunks, and where muscle nerves branch off, growth cones were larger and more complex and turned more often. In straighter stretches they were smaller and simpler. The work located guidance decisions at particular places and gave later molecular studies a set of choice points to examine.

✓ verbatim Tosney & Landmesser, J. Neurosci. 5:2345 (1985); Tosney & Landmesser, J. Neurosci. 5:2336 (1985).
Test yourself
Question 1 of 8 Score 0 / 0

Cells at the ventral midline of the embryonic spinal cord secrete a protein that spreads through the tissue and draws commissural axons toward the midline from a distance. Which class of guidance cue does this describe?
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

  • Laminin alpha 2 LG4-5 domain pair — a part of the basement membrane protein that supports axon growth — RCSB PDB 1DYK · coordinates CC0 · Mouse laminin alpha 2 chain LG4-LG5 domains, X-ray crystal structure (Tisi, Talts, Timpl & Hohenester, EMBO J., 2000)
  • N-cadherin, two-domain fragment — a calcium-dependent adhesion molecule on growth cones — RCSB PDB 1NCJ · coordinates CC0 · Mouse N-cadherin extracellular domains 1 and 2, X-ray crystal structure (Tamura, Shan, Hendrickson, Colman & Shapiro, Neuron, 1998)
  • NCAM Ig1-3 fragment — a vertebrate relative of fly Fasciclin II — RCSB PDB 1QZ1 · coordinates CC0 · Rat NCAM immunoglobulin-like domains 1 to 3, X-ray crystal structure (Soroka et al., Structure, 2003)
  • Integrin alphaVbeta3 extracellular segment — the heterodimer shape of matrix receptors — RCSB PDB 1JV2 · coordinates CC0 · Human integrin alphaV beta3 ectodomain, X-ray crystal structure (Xiong et al., Science, 2001)

Scientist profiles & quotes

  • David Bentley — Bentley & Caudy, Nature 304:62 (1983); Caudy & Bentley, J. Neurosci. 6:1781 (1986).
  • Corey Goodman — Raper, Bastiani & Goodman, J. Neurosci. 3:31 (1983) and 4:2329 (1984); Lin & Goodman, Neuron 13:507 (1994); Fambrough & Goodman, Cell 87:1049 (1996).
  • Lynn Landmesser — Tosney & Landmesser, J. Neurosci. 5:2345 (1985); Tosney & Landmesser, J. Neurosci. 5:2336 (1985).

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

Animated explainer

A Long Route in Short Legs · 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 growing nerve cell sends out a long arm that has to find the right place, like a hiker on a long trail. Along the way there are signposts: some say come here, some say keep away, some you touch and some you smell from far off. The first hikers mark the trail, and later hikers hold hands with the ones who went before. If you pull up a signpost, the first hikers get lost.
High school

Axons grow by a moving tip called the growth cone, and it steers by reading signals called guidance cues. Some cues are stuck to cells or to the material between them and act on touch; others are released and act from a distance. Each kind can attract or repel. Long routes are broken into short legs. The first axons, called pioneers, steer by special cells along the way called guideposts. In grasshopper embryos, killing one guidepost cell made the pioneers lose their way. Later axons often follow the pioneers by sticking to them, and choose which bundle to follow by recognizing molecules on its surface.
College

Guidance cues fall into four classes: contact attraction (laminin, Ig-superfamily adhesion molecules such as L1 and Fasciclin II), contact repulsion (ephrins), long-range attraction (netrin-1 from the floor plate) and long-range repulsion (Slit, secreted semaphorins). Pioneer axons cross territory with no axons ahead of them, using guidepost cells and intermediate targets that divide the route into legs; killing the Cx1 guidepost disrupts the route of grasshopper Ti1 pioneers. Followers fasciculate onto pioneers, and in the labeled pathway hypothesis each bundle carries surface labels that particular growth cones recognize. In Drosophila, loss of the homophilic adhesion molecule Fasciclin II lets bundled axons separate, while extra Fasciclin II keeps motor axons bundled; the secreted protein Beat opposes this adhesion so motor axons can leave their nerve. Substrates act through cadherins, Ig-CAMs and integrins, and growth is fastest at intermediate adhesion. A cue's sign is set by the receptors and internal state of the reading cell.
Graduate

The four-class scheme describes how a cue behaves in an assay, not a fixed property of the molecule: netrin-1 can act locally when bound to matrix or cell surfaces, and the sign of the response depends on receptor complement (DCC alone versus DCC with UNC5) and on second messengers, with cAMP and calcium levels converting attraction to repulsion in Xenopus growth cones in culture. Genetic analysis in Drosophila supports a dose-sensitive balance at choice points: Fasciclin II loss and gain give opposite fasciculation phenotypes, and reducing Fasciclin II suppresses the bypass phenotype of beat mutants, which places Beat upstream of, or in opposition to, Fasciclin II-mediated axon-axon adhesion rather than in a steering pathway toward the muscle. Adhesion and speed are related non-monotonically, as motor-clutch models predict. An open question is how a growth cone integrates several cue systems acting at the same choice point, each at a different range and sign, into one decision that it is costly to reverse.
A better experiment they did not do

Killing the P neurons showed that G needed them, but not why. Perhaps the P axons carried a label that told G which way to go, or perhaps they simply offered the only usable surface on a route chosen for other reasons. The ablation alone could not tell these apart. An experiment that moved the A/P axons to a new place in otherwise intact tissue, and asked whether G follows them there, would have decided it.

— Raper, Bastiani & Goodman, Pathfinding by neuronal growth cones in grasshopper embryos IV, J. Neurosci. 4:2329 (1984).
Why grasshopper neurons light up

Antibodies raised against horseradish peroxidase, a plant enzyme, turned out to stain the neurons of insects, including the grasshopper and the fruit fly, because the antibodies recognize a sugar group carried by insect neuronal surface proteins. Lily and Yuh Nung Jan reported the staining in 1982; the sugar basis was worked out later. The accident gave insect embryologists a stain that marks neurons in whole embryos, and the Jans themselves used it to follow how the fly and grasshopper nervous systems form.

— Jan & Jan, Antibodies to horseradish peroxidase as specific neuronal markers in Drosophila and in grasshopper embryos, Proc. Natl. Acad. Sci. USA 79:2700 (1982).

 ↑ Top