Radial Migration: Climbing the Glial Ladder

A newborn cortical neuron is about ten micrometres across, and the wall it has to cross can be many times that thick. It makes the trip with no legs, pulling its own nucleus forward one stroke at a time.

You have seen in the lesson on birth order and layers that a neuron’s birth date predicts its layer: early-born neurons settle deep, and later-born neurons settle in more superficial layers. That rule needs travel. Excitatory neurons of the cortex are born at the inner surface of the wall, in the ventricular zone and subventricular zone, beside the fluid-filled ventricle. Their layers lie near the outer, pial surface. In between is the intermediate zone, where axons run. In a mouse the wall is thin at first; in a primate late in gestation it is far thicker. So every neuron faces the same physical problem: how does a cell move its bulky nucleus outward through a crowded wall, over a distance many times its own length?

This lesson answers in three parts: the route, the ladder, and the engine.

The idea: two ways to go up

Professor Deppmann frames this as “Modes of radial migration: translocation versus locomotion.”

In 2001 Bagirathy Nadarajah, Alan Pearlman, Rachel Wong and colleagues filmed cells in acute slices of embryonic mouse cortex by time-lapse imaging. Early in corticogenesis, many cells had a long leading process already attached near the pial surface. The process shortened and the cell body moved up within it, fairly smoothly. They called this somal translocation. Picture a person on a rope that is already tied to the ceiling, simply reeling in.

Other cells had a shorter leading process that ended freely, and their cell bodies advanced in jumps. They called this locomotion, and from the cells’ shape they inferred that it was guided by radial glia. Locomotion becomes the main mode later, when the wall is thick. They also saw some locomoting cells switch to a final translocation once the leading process reached the marginal zone at the top of the cortical plate.

The route of a late-born neuron is not a straight climb. In 2004 Stephen Noctor, Arnold Kriegstein and colleagues labelled progenitors in embryonic rat cortex with a GFP retrovirus and filmed their daughters in slices. Newborn neurons first moved to the subventricular zone, then became multipolar, with several short processes, and paused there or in the lower intermediate zone, in Professor Deppmann’s summary for up to about a day. Most then moved back toward the ventricle for a time before turning bipolar and migrating radially to the cortical plate. Only a movie could show it.

The glial ladder: from an inference to a movie

The fibres the neurons climb belong to radial glia. As the lesson on radial glia explains, these same cells are the main progenitors of the cortex. Each one keeps a process running from the ventricle to the pia, so the wall is strung with parallel fibres.

Pasko Rakic proposed in 1972, for the cerebral cortex, that these fibres are the ladder; he had made the same case for the cerebellum in 1971. He rebuilt young neurons from serial electron micrographs of fetal rhesus monkey cortex and found them pressed along radial glial fibres at every depth of the intermediate zone. That was fixed tissue. It showed contact, and it suggested a climb, but it could not show which cell moved, or how.

In 1987 James Edmondson and Mary Hatten supplied the movie, in a different animal and a different brain region. They purified granule neurons and astroglia from the cerebella of early postnatal mice, recombined them on a coverslip, and filmed them. Neurons moved along glial processes in a stop-and-go rhythm, with the cell body closely apposed to the fibre and a leading tip that extended and retracted out of step with the soma. A later antibody stain confirmed that the guides were glial. The experiment showed that a neuron can migrate on a glial fibre and how it moves when it does. It did not show that glia are required in the living brain, or what tells a neuron to get off.

The “two stroke” cycle: how the nucleus is pulled

Here is the engine. Professor Deppmann calls nucleokinesis, the forward movement of the nucleus, a “two stroke” cycle.

1. Stroke one. The leading process extends. A swelling forms in it, ahead of the nucleus, and the centrosome and Golgi move forward into the swelling.
2. Stroke two. The nucleus is pulled forward into the swelling. The trailing part of the cell follows.

Then the cycle repeats. The analogy is an inchworm: the front reaches and grips, then the back catches up.

Bruce Schaar and Susan McConnell watched this cycle in neurons from the anterior subventricular zone of newborn rats, migrating in a gel. The swelling predicted where the nucleus would go next. Electron micrographs of cells fixed mid-cycle showed the centrosome ahead of the nucleus and bundles of microtubules wrapped around the nucleus like a cage, with the microtubules running forward toward the centrosome.

That cage is the track. Cytoplasmic dynein is a motor that walks toward the minus ends of microtubules, which sit at the centrosome. Anchored around the nucleus and pulling on that cage, dynein is thought to draw the nucleus toward the centrosome. A second force comes from actin and myosin II, the contractile system of the lesson on integrins and Rho GTPases, where Rho and its kinase ROCK switch on myosin. When Schaar and McConnell added blebbistatin, which blocks myosin II, leading processes could keep growing while the cell bodies stopped. Whether myosin squeezes from behind the nucleus or pulls from the front is still argued. Different preparations put actomyosin activity in different places, and in some cerebellar granule neurons the nucleus even passes the centrosome. The force distribution likely depends on cell type and stage.

A movie gives speed directly. If each stroke moves the nucleus a fixed step:

mean speed = step per cycle × cycles per hour

Motor-driven nuclear movement also occurs earlier, in the progenitors themselves. In “Interkinetic nuclear migration” the nucleus of a radial glial cell moves away from the ventricle in G1 and back to it in G2, dividing at the ventricular surface. In embryonic rat cortex the kinesin Kif1a, a plus-end motor, carries it basally, and dynein carries it apically. That is a contrast worth keeping: progenitors use two motors in two directions, while a migrating neuron uses dynein to move its nucleus toward a centrosome that keeps moving forward.

Design problem

You image a migrating cortical neuron in a slice and see that its leading process keeps extending but the nucleus stays put. Name two molecular defects that could cause this and one experiment that tells them apart.

”The mutants” and “Human lissencephaly”

Professor Deppmann’s list of “The mutants” runs from LIS1, NudE and Ndel, doublecortin, Cdk5 and dynein to myosin II, ROCK and many more. Most of the names fit one of the two strokes.

LIS1. In 1993 Orly Reiner, David Ledbetter, William Dobyns and colleagues isolated LIS1 from chromosome band 17p13.3 and showed it was deleted in Miller-Dieker lissencephaly. Lissencephaly means a cortex with reduced or absent folding; in the classic form the cortex is also abnormally thick and poorly layered, because neurons failed to reach their places. Losing one copy of LIS1 is enough. The clue to its job came from a fungus: in 1995 the laboratory of N. Ronald Morris found that nudF, a gene needed to spread nuclei along a filament of Aspergillus, encodes a protein similar to LIS1. Other genes from the same screen encode parts of the dynein system. By 2000 several groups had linked mammalian LIS1 to dynein and found LIS1, NDEL1 and dynein together in the nervous system. LIS1 is now understood as a regulator of dynein, not the motor itself.

Doublecortin. In 1998 two groups (Gleeson, Walsh and colleagues; des Portes, Chelly and colleagues) identified DCX on the X chromosome. It encodes a protein that binds and stabilizes microtubules. Boys with a mutation typically have lissencephaly. Girls with one mutant copy are mosaics, because each cell inactivates one X: neurons using the normal copy migrate, and some using the mutant copy stall. The stalled neurons form a band of grey matter in the white matter below the cortex, called subcortical band heterotopia or “double cortex”.

Cdk5. Cdk5 is a kinase of the cell-cycle family that, in neurons, is activated by p35 instead of a cyclin. Mice lacking Cdk5 or p35 have disordered cortical layers. Among its targets are NDEL1 and doublecortin, so Cdk5 sits upstream of the same nuclear machinery.

These patients and mice show what the motor system is for, but each gene also acts at other stages. LIS1 knockdown in embryonic rat cortex blocks the nuclear movement of progenitors, the multipolar-to-bipolar change and the movement of the soma, so a single defect can stall cells at more than one step.