Netrin and DCC: One Cue, Two Answers

Put a scrap of floor plate from an embryonic rat spinal cord in a gel near a scrap of dorsal spinal cord, and axons grow out of the dorsal scrap toward the floor plate, though the two never touch.

The floor plate is a strip of cells along the ventral midline of the embryonic spinal cord. Commissural neurons sit in the dorsal spinal cord, and their axons grow ventrally to the floor plate and cross it to the other side. The gel result says the floor plate gives off something that travels. This lesson asks what that something is, how the growth cone (the moving, sensing tip of a growing axon; see the lesson on the growth cone) reads it, and why the same molecule can pull one axon in and push another away.

Classic Experiment: Floor Plate Chemoattraction

In 1988 Marc Tessier-Lavigne, Marysia Placzek, Andrew Lumsden, Jane Dodd and Thomas Jessell embedded pieces of rat embryonic spinal cord in a three-dimensional gel of collagen. On its own, an explant of dorsal spinal cord sent out axons in no preferred direction. With a floor plate explant nearby, bundles of commissural axons grew out of the dorsal explant toward the floor plate. Because the explants were apart, the effect had to cross the gel. A cue that acts this way is a chemoattractant, and steering by it is chemotropism.

That fits one of the four classes of cue you met in the lesson on guidance cues and pioneers: long-range attraction. It does not tell you what the molecule is. Picture smelling bread from down the street. You know there is a bakery, and you know which way to walk, but not the recipe.

Purifying the cue: chick brains and a worm gene

To purify a protein you need a lot of tissue, and floor plates are tiny. The answer, from Tessier-Lavigne’s laboratory in 1994 with Tito Serafini as first author and Timothy Kennedy as second, was to keep the rat assay as a detector and to look for the same activity in a larger source. Membranes from embryonic chick brain had it. Thousands of brains went into the work. The extract was separated over several chromatography columns, and after each step every fraction went back onto rat spinal-cord explants. The axons reported which fractions to keep.

Two proteins came through. Their sequences were related to laminin, a protein of the extracellular matrix, and to UNC-6, the product of a worm gene. The authors named them netrin-1 and netrin-2. Each promoted commissural outgrowth.

Outgrowth is not direction. A companion paper by Kennedy, Serafini, José de la Torre and Tessier-Lavigne supplied direction. Floor plate cells make netrin-1 RNA, and cells engineered to secrete netrin attracted commissural axons at a distance in the gel.

The worm had got there first. In 1990 Edward Hedgecock, Joseph Culotti and David Hall followed cells and pioneer axons (the first axons to lay down a route) that migrate around the body of C. elegans, between its ventral and dorsal sides. Mutants of unc-6 went wrong in both directions. Mutants of unc-40 failed mainly in movement toward the ventral side, and mutants of unc-5 mainly in movement away. UNC-6, it turned out in 1992, is a laminin-related protein. So before vertebrate netrin had a name, worm genetics said: one cue, two readers, two directions.

The receptors: one cue, two answers

UNC-5 was identified in 1992 as a transmembrane protein, and in 1993 Michel Hamelin and colleagues in Culotti’s laboratory forced worm touch neurons to make it. These neurons do not normally carry UNC-5 and do not grow dorsally. With UNC-5 their axons grew dorsally, away from the ventral UNC-6, and this still needed UNC-6. Changing the receptor reversed the answer to an unchanged cue.

UNC-40 turned out to resemble a human protein with an odd name: DCC, for Deleted in Colorectal Cancer. The gene had been found in 1990 on a stretch of chromosome 18 often lost in colorectal tumours and studied as a candidate tumour suppressor (see the lesson on oncogenes). In 1996 Kazuko Keino-Masu, Masayuki Masu and colleagues in Tessier-Lavigne’s laboratory showed that in rat embryos DCC is on commissural axons, that netrin-1 binds cultured cells displaying DCC, and that an antibody to DCC selectively blocks netrin-evoked outgrowth from spinal explants. In the same journal issue, Shirley Chan and colleagues showed that worm UNC-40 is needed in the moving cells themselves.

Other systems point the same way. The fly Unc5 receptor is a repulsive netrin receptor (Keleman and Dickson, Neuron 2001), and a 2014 crystal structure showed that one netrin can bind DCC at one site while UNC5A can take DCC’s place at a second, generic site, which the authors proposed switches the response from attraction to repulsion (Finci and colleagues). A 1999 frog-neuron paper that had reported this switch with chimeric receptors (Hong and colleagues) was retracted in 2023, so this lesson does not rely on it. The working rule is that DCC alone attracts and UNC5 receptors add repulsion, but DCC-family receptors also take part in some repulsion, and netrin has other receptors and partners in some tissues, so treat the rule as a starting point.

Attraction and repulsion can flip

Receptors are one layer. The growth cone’s internal state is a second. In the turning assay from Poo’s laboratory, a fine pipette puffs a cue near one cultured growth cone, and time-lapse imaging measures the turn. In 1997 Guo-li Ming, Hong-jun Song and colleagues found that growth cones of frog (Xenopus) spinal neurons turn toward netrin-1. When they blocked cAMP signalling, with a competing cAMP analogue or an inhibitor of protein kinase A (PKA), the same growth cones turned away. An antibody to DCC stopped both turns. The receptor was the same, and the cell’s cAMP chose the direction. (For how cAMP activates PKA, see the lesson on cAMP and PKA.)

Calcium is part of the readout. In 2003 Makoto Nishiyama and colleagues reported that the ratio of cAMP to cGMP in these growth cones sets the direction of the netrin-1 turn by changing the activity of L-type calcium channels (see the lesson on calcium signaling).

The eye shows why a flip might be useful. Retinal axons run along the inner surface of the retina to the optic nerve head, the exit, where netrin-1 is concentrated, and netrin attracts them there. In 1999 Vivian Höpker and colleagues found that laminin-1 turns netrin’s attraction into repulsion in frog retinal growth cones, and that laminin-1 lowers their cAMP. Laminin-1 lies on the retinal surface. Their model: where laminin and netrin overlap, netrin repels, and this pushes axons down into the exit, where there is netrin without laminin. Netrin is attractive at one point and repellent at the next, with cAMP deciding.

Long-Range vs Short-Range Cues

A growth cone steering in a gradient compares the cue on one side with the cue on the other, a few micrometres apart. For steady diffusion from a small source in three dimensions, concentration falls as 1/r, and the fractional difference across a growth cone is roughly its width over its distance from the source:

fractional difference ≈ w / r (for C = k / r; w = growth cone width, r = distance)

A 10 micrometre growth cone 100 micrometres away sees about a 10 percent difference. A millimetre away it sees 1 percent. Receptor binding is also noisy, because ligand molecules arrive and leave at random, so one snapshot of occupied receptors on each side cannot be trusted. Yet in 2004 Rosoff and colleagues, with gradients they could control, found that axons are extremely sensitive to gradients, but only within a narrow range of concentrations. In 2002 Ming and colleagues found part of the answer in frog growth cones: in a steadily rising background of netrin-1 they lost sensitivity and then regained it. A detector that resets its baseline can keep reading a gradient as it climbs it.

Design problem

How would you experimentally test whether Netrin acts as a long-range diffusible gradient versus a short-range contact cue? Design a simple in vitro experiment.