Harrison's Living Axon and the Growth Cone

In 1907 Ross Harrison put a scrap of frog embryo into a drop of clotted lymph, and over the following days he watched nerve fibres grow out of it.

It settled an argument that fixed tissue could not. An axon is the long fibre a nerve cell sends out to reach its partner, and by the late 1800s two pictures of how it formed were in play. In one, the axon grows out from a single nerve cell, as an extension of that cell. In the other, the nerve forms in place along its route, from a chain of cells or from strands of living material (one version, associated with Victor Hensen, called them protoplasmic bridges) already joining the nerve centre to its destination.

The idea: a living fibre decides what a fixed one cannot

Wilhelm His and Santiago Ramón y Cajal argued for the first picture from dead tissue. Cajal used Golgi’s silver chromate stain, which blackens only a few cells in a sample but blackens each of them almost entirely. In the spinal cords of chick embryos a few days old, he found young nerve fibres of different lengths, each ending in a swollen club with spikes and fine projections. In 1890 he named it the growth cone and argued that it was the front of an axon on the move.

The trouble was that every one of his specimens was fixed. A row of fibres of different lengths can be read as one fibre growing, or as fibres that were always there becoming visible as they mature. More stained sections would not settle it. Someone had to watch a fibre while it was alive.

Harrison’s preparation was built to make the answer unambiguous. He removed a piece of the neural tube (the embryo’s first nervous system, a tube of cells along its back) from frog embryos at a stage when no nerve fibres could yet be seen. He placed it in a drop of fresh lymph from an adult frog on a coverslip, inverted the coverslip over a hollow slide and sealed it. The lymph clotted into a soft mesh that held the tissue. This is an explant: a piece of tissue kept alive outside the body.

Fibres then grew out of the tissue into the clot, and Harrison saw their swollen ends change shape. Each feature of the design closed off an alternative. The tissue was taken before any fibres existed, so these were not old fibres exposed by dissection. The clot contained no living chain of cells for a nerve to form from. And in some cultures a fibre could be seen arising from a single isolated cell. An axon, he concluded, is drawn out from one cell by movement at its tip. He reported this in 1907, and a longer paper in 1910 drew mostly on later preparations he judged more convincing.

Harrison also left a question open. His fibres grew into an unorganized clot, with nothing to point them anywhere. In the embryo an axon has to reach a particular partner. Think of an axon growing to its target as a driver navigating city streets: the car has to move before it can be steered, and the driver has to read the signs. The growth cone is the car and the driver at once.

Growth cone is both a sensory and motor structure

A growth cone in culture is a flat fan at the end of the axon. Its parts divide the labour.

• The central domain sits where the axon shaft widens into the fan. It is packed with microtubules, hollow tubes about 25 nanometres wide built from dimers of alpha and beta tubulin.
• The peripheral domain is the fan itself, built of actin filaments, which are much thinner (roughly 6 to 10 nanometres). In filopodia, the thin spikes at the edge, actin filaments are bundled in parallel. In lamellipodia, the sheets between the spikes, they form a branched mesh.
• The surface carries receptors for guidance cues and adhesion molecules that grip the substrate or other cells.

The filopodia are both feelers and feet. They carry receptors and adhesion molecules out ahead of the growth cone, sample what is there, and attach. The growth cone is both a sensory and motor structure: the same tip detects the environment and moves through it.

Harrison’s next question was about the motor side. In 1914 he asked whether cells need a solid surface or can move through fluid alone. He gave cultures three kinds of support: clotted plasma, smooth glass, and threads of spider web. He also suspended some pieces of tissue in large drops so that they touched nothing. Cells from frog embryos wandered along the web threads, but cells in unsupported drops generally did not move. Pieces of forebrain from five-day chick embryos produced nerve fibres in all four preparations in clotted plasma, and none on spider web or in unsupported drops. Contact with a solid was needed, but not every solid would do.

So: “Do growth cones like adhesive or non-adhesive substrate?” Adhesive. Later culture experiments put growth cones on dishes patterned with surfaces of different adhesiveness. In Paul Letourneau’s 1975 study with chick sensory neurons, growth cones tended to follow the more adhesive surface. “Do growth cones like more adhesive or less adhesive substrate?” More, when given a choice. Preference is not speed. In culture, how adhesive a surface is turns out to be a poor predictor of how fast axons grow on it (Lemmon and colleagues, 1992).

How do growth cones grow? The clutch hypothesis

Inside the fan, actin is never still. New subunits join the filaments at the end that faces the leading edge, while motor proteins (myosin II) pull the whole network back toward the centre, where filaments are taken apart. This backward movement is called retrograde flow. The network treadmills.

Paul Forscher and Stephen Smith saw this in growth cones of Aplysia, a sea slug with large, flat growth cones. In 1988 they added a cytochalasin, a drug that stops actin subunits adding at the leading edge. A band free of actin opened at the front and widened, while the remaining network kept moving back. Microtubules, normally kept in the centre, then pushed forward into the emptied band. The actin had been flowing backward all along, and its mesh had been holding the microtubules back.

If actin is always flowing back, how does the edge ever move forward? Tim Mitchison and Marc Kirschner proposed an answer in 1988: the clutch hypothesis. Think of the actin network as a running engine and the substrate as the road. When the clutch is disengaged, the engine runs and the car stays put: actin is added at the front and carried back at the same rate, and the edge goes nowhere. When adhesion molecules link the network to the substrate, the clutch engages. Retrograde flow slows, and the same polymerization now pushes the edge forward.

rate of advance = rate of actin addition at the edge − rate of retrograde flow

In 1995 Forscher’s laboratory reported that where an Aplysia growth cone made a firm contact, flow slowed behind it and the growth cone advanced that way. Some of the most detailed measurements of the clutch’s molecular linkage come from non-neuronal migrating cells (reviewed by Case and Waterman, 2015), not from growth cones. In many cells the clutch’s grip on the surface runs through integrins, receptors for the extracellular matrix, while the Rho-family GTPases Rho, Rac and Cdc42 control where actin is built and how hard myosin pulls (see that lesson); like every small GTPase, each is a switch that is on when bound to GTP (that lesson). At the tips of filopodia, Ena/VASP proteins keep filament ends from being capped, so the spike keeps lengthening.

Actin makes the advance; microtubules make it permanent. Once the edge has moved and the actin ahead has cleared, microtubules extend into the new territory, the fan behind them narrows, and that stretch becomes axon shaft.

How do growth cones turn?

If advance is a balance at the edge, turning is a difference in that balance between the two sides. Experiments that release a drug onto one side of a growth cone give a simple set of results. Remove actin on one side and the growth cone turns away from that side. Stabilize microtubules on one side and it turns toward that side. Destabilize them on one side and it turns away. A guidance cue works the same way: it shifts actin assembly, flow or adhesion on the side nearer the source, and microtubules then follow the side that is gaining.

The best-known assay for this is a pipette. In 1979 Gundersen and Barrett placed a micropipette releasing nerve growth factor (NGF) near growing axons of chick sensory neurons. The growth cones turned toward the pipette, as they did within about 20 minutes (Gundersen and Barrett, 1980, J Cell Biol 87:546).

Where is that decision made? “Axons can respond to cues without a cell body.” In 2001 Campbell and Holt cut the axons of retinal neurons from Xenopus frog embryos, leaving growth cones with no cell body attached, and the isolated growth cones still turned in response to guidance cues. They also needed to make protein to do it: blocking protein synthesis in the growth cone itself blocked the turning. The cell sends messenger RNA down the axon ahead of time, and the growth cone translates it at the tip. In a long axon, the cell body can be far too distant to answer in time; the tip decides.

Design problem

A growth cone from a chick sensory neuron turns toward a pipette releasing NGF. Design manipulations to test whether turning needs (a) actin dynamics on the near side, (b) microtubules on the near side, and (c) the cell body. Say what result each would give if the part is needed.