Reelin and the Reeler Mouse

In 1951 a geneticist named a mouse mutant for the way it walks: it reels from side to side. Its cerebellum is small, and the layers of its cerebral cortex are disorganized and roughly in the reverse of their normal order.

A walking problem and a cortex built in the wrong order turned out to be two faces of one missing protein. This lesson asks how a mouse that falls over led to a signaling pathway, and what that pathway tells a migrating neuron. In radial migration you followed a neuron climbing a radial glial fiber from the ventricular zone toward the surface. Here the question is where it stops, and who decides.

Naturally occurring mutations: a mouse named for how it walks

Before anyone could knock out a gene on purpose, geneticists relied on mutants that arose on their own in mouse colonies. A breeder notices an odd animal, breeds it, and keeps the line. Several of the most useful brain mutants were named by their gait: reeler, weaver, staggerer. The name records what the mouse does, not what is wrong inside it.

Douglas Falconer described and named reeler in 1951. A mutant is a handle: you can breed it, map it and dissect it long before you know its gene. Reeler mice have a cerebellum that is small and poorly folded, with Purkinje cells, the large output neurons of the cerebellum, left in clumps instead of a neat single layer. That alone could explain the reeling. The cortex was the deeper puzzle.

New neurons push past the old neurons, normally

Recall how the cortex is built (see birth order and layers). The first neurons to leave the ventricular zone form a thin sheet just under the surface, the preplate. The next neurons to arrive move into the middle of that sheet and split it in two: an upper marginal zone, which holds the Cajal-Retzius cells, and a lower subplate. The neurons between them form the cortical plate, which becomes the six layers. Each new wave climbs past the wave before it and settles just under the marginal zone, so early-born neurons end up deep and late-born neurons end up superficial. Professor Deppmann puts it in one line: “New neurons push past the old neurons…normally.”

The “normally” is where reeler comes in. Was reeler making the wrong neurons, or putting the right ones in the wrong place? Verne Caviness and Richard Sidman answered with birthdating. A pulse of tritiated thymidine, a radioactive building block of DNA, labels cells that are copying DNA. A neuron that divides for the last time soon after the pulse keeps a heavy label for life, so its position in the adult tells you where neurons born on that day settle.

In their 1973 paper, the same classes of neurons were born on the same schedule in reeler as in normal mice. What changed was their order. In reeler, early-born neurons tended to sit superficially and late-born neurons deep, a roughly inverted pattern, and the layers were disorganized as well. Later work showed why it starts wrong: in reeler the preplate fails to split. The first cortical plate neurons pile up beneath it instead of entering it, and every later wave stops beneath the neurons already there instead of passing them.

So reeler is a defect of placement. The neurons are made. They cannot get past their elders.

Where does Reelin come from? Cajal-Retzius cells

In 1995 two lines of work began to converge on the same protein. A team led by Gabriella D’Arcangelo and Tom Curran cloned the gene disrupted in reeler, starting from a mouse in which an inserted transgene had broken it. The gene encodes a very large protein that resembles extracellular matrix proteins, and it is made in the developing brain while neurons migrate. They named it Reelin. A separate team, Hirotsune and colleagues, reported the reeler gene in the same year.

The second route used an antibody. Masaharu Ogawa, Katsuhiko Mikoshiba and colleagues made an antibody called CR-50 that stained Cajal-Retzius cells in the marginal zone. Added to cultures in which embryonic cortical cells reassemble, it made normal tissue organize like reeler tissue. In 1997 the two groups together showed that CR-50 recognizes Reelin and that Reelin is a secreted glycoprotein.

That fixes the geometry. Reelin is made by cells at the top of the cortex, the very place where each new wave of neurons should stop. Cajal-Retzius cells are early neurons, and most arrive in the marginal zone from a few sites at the edges of the developing cortex.

The Reelin pathway: four mutants, one phenotype

How does reelin work? A secreted protein outside the cell needs a receptor and a relay inside. The pieces came from mutants that looked the same.

Two more mouse mutants, scrambler and yotari, had a reeler-like brain but a normal Reelin gene. In 1997 both turned out to be mutations in Dab1 (Disabled-1), and in the same issue of Nature Jonathan Cooper’s laboratory reported that a targeted knockout of Dab1 gave the same picture. Dab1 is an adaptor: a protein with no enzyme activity of its own that links one protein to others (see adaptors and scaffolds). Then in 1999 Joachim Herz’s laboratory showed that mice lacking both of two lipoprotein receptors, ApoER2 and VLDLR, also look like reeler, while either single knockout is milder.

Why do identical phenotypes put genes in one pathway? Think of a relay race. Remove any runner and the baton never arrives. The race fails the same way whichever runner is missing. When removing Reelin, Dab1 or both receptors gives the same failure, the simplest reading is that they are runners in one race. Matching phenotypes alone do not tell you the order, so it came from biochemistry. Brains lacking Reelin had more Dab1 protein than normal but less of it phosphorylated, and adding Reelin to cultured neurons raised Dab1 tyrosine phosphorylation within minutes. Two papers in one 1999 issue of Neuron, from the Curran and Herz laboratories, showed that Reelin binds ApoER2 and VLDLR directly.

The pathway reads like this:

Reelin (from Cajal-Retzius cells) → binds ApoER2 and VLDLR → Fyn and Src phosphorylate Dab1 on tyrosine → phospho-Dab1 recruits effectors → phospho-Dab1 is ubiquitinated through Cullin 5 and degraded

Dab1 sits on the receptor tails through its PTB domain, which recognizes an NPxY motif (see protein interaction modules). Mice lacking both Fyn and Src have a reeler-like cortex, which places these Src-family kinases in the relay. The last step is the off switch. Phosphorylated Dab1 is marked by a ubiquitin ligase built on Cullin 5 and destroyed (see ubiquitin and proteolysis and turning signals off). That helps explain the surplus Dab1 in reeler: with no Reelin, Dab1 is poorly phosphorylated and so degraded less efficiently. When Cooper’s laboratory removed Cullin 5 from migrating mouse neurons, active Dab1 built up and the neurons migrated too far, piling up at the top of the cortical plate.

In humans, rare recessive RELN mutations cause lissencephaly with cerebellar hypoplasia, and in mice Reelin also modulates synaptic plasticity in the adult hippocampus.

Reelin: chemoattractant or stop signal?

Professor Deppmann’s question is the one the field still argues: “Reelin: chemoattractant or stop signal?” Neurons pile up wrongly without it, but a missing signal tells you only that it is needed, not what it says.

The stop-signal view: Reelin in the marginal zone tells an arriving neuron to stop. In a related detachment view, Reelin tells the neuron to let go of its radial glial fiber, so it can finish its last step and leave room for the next wave to pass. Without Reelin, neurons never get the order and newcomers cannot pass.

The attractant view: Reelin draws the leading process of a neuron toward itself. In 2010 Kazunori Nakajima’s laboratory expressed Reelin ectopically in the developing mouse cortex. The leading processes of migrating neurons gathered in the Reelin-rich spot, cell bodies collected around it, and later-born neurons passed earlier-born ones toward the center, building a small inside-out arrangement around the ectopic source.

A third line of work splits the question by stage. In mouse cortex, Reelin signaling through Dab1 and the small GTPase Rap1 has been linked to orienting multipolar neurons, to adhesion through N-cadherin, and to the final glia-independent step, called somal translocation, in which the neuron anchors its leading process in the marginal zone and pulls its cell body up. That last step needs integrin adhesion to the marginal zone. On this view Reelin is less a command than a switch that changes adhesion at specific steps, and the effect depends on the stage the neuron is in.

The answer is still open, partly because the experiments point different ways. Ectopic Reelin made in the ventricular zone of reeler mice rescued some defects but not all, which says Reelin is not simply a positional cue. Focal Reelin organized neurons around it, which says its location can matter. The models are not all exclusive, and you should leave this lesson holding more than one.

Design problem

Make Professor Deppmann’s question testable. Design an experiment, in cortical slices or in living mouse embryos, that distinguishes Reelin as a chemoattractant (neurons read a gradient and move toward the source) from Reelin as a stop or detachment signal (neurons stop or let go of glia wherever they meet it).