Neurulation and Neural Tube Defects

In a human embryo, the tissue that becomes the brain and spinal cord starts as a flat sheet of cells, and by the end of about the fourth week after fertilization it has rolled into a closed tube. That is often before a woman knows she is pregnant.

So the event this lesson is about happens early, quickly and out of sight. A plate of tissue on the embryo’s back has to thicken, narrow, fold and seal itself shut. If one step fails, part of the nervous system stays open for the rest of pregnancy. The professor’s question for this stage is plain: “What happens if this doesn’t close?” To answer it you need to know how closing works, which experiments showed it, and why a vitamin taken before conception changes the odds.

Neurulation: a flat plate rolls into a tube

Start with “Neurulation” in one picture, the “Neural Plate - D/V” top view, looking down on the embryo’s back. The ectoderm is the outermost of the embryo’s three germ layers. After neural induction, a band of it along the midline becomes the neural plate, the sheet that will form the brain and spinal cord. The ectoderm on either side stays epidermis, the future skin. Beneath the midline runs the notochord, a rod of mesoderm.

Primary neurulation turns the plate into a tube in overlapping steps, worked out mostly in chick and mouse embryos:

1. Shaping. The plate cells grow taller, so the plate thickens. At the same time the plate narrows from side to side and lengthens from head to tail.
2. Bending. The plate creases along hinge points. In chick and mouse a medial hinge point lies over the notochord, and paired dorsolateral hinge points form at some levels of the body. Cells at a hinge shrink their apical face, the side facing away from the embryo, and become wedge-shaped. Actin and myosin drive that squeeze, with Rho kinase helping in chick, as that lesson describes for other cells. The edges rise as the neural folds.
3. Meeting and fusion. The folds move toward the midline, meet and fuse.
4. Separation. The tube detaches from the epidermis, which seals over it.

Neural crest cells arise at the border where the neural plate meets the epidermis. They leave the folds around the time of fusion and form much of the peripheral nervous system; how they leave is a later lesson.

Closure does not happen all at once. In the mouse, fusion begins at several separate sites and spreads from each like a zipper, and the last openings, called neuropores, close at the ends. Whether human closure starts from the same set of sites is debated.

The tail end is built another way. There, cells of the tail bud gather into a solid rod that then hollows out and joins the tube made above it. This is secondary neurulation, and in humans it is thought to form the lowest sacral and tail-end segments of the cord.

Narrowing the plate: convergent extension

Why does the plate have to narrow? Think of two people on opposite sides of a wide table trying to shake hands. They can lean, but their arms have a fixed reach. The folds are the same: they can rise and move inward only so far. If they start too far apart, they never meet.

The narrowing comes from convergent extension. Cells in the plate push in between their neighbours on either side, a move called intercalation. Each such step makes a row one cell narrower and longer. Repeated across the sheet, many small exchanges turn a short, wide plate into a long, thin one.

For that to work, the cells need to agree on which way is side to side. Planar cell polarity (PCP) is a sense of direction within the plane of a sheet, set by core proteins including Frizzled, Dishevelled, Vangl and Prickle that gather on opposite sides of each cell. In vertebrates this is the branch of Wnt signaling that runs through Dishevelled without beta-catenin and reaches the cytoskeleton, partly through Rho; that lesson covers the branches.

John Wallingford and Richard Harland tested the idea in embryos of the frog Xenopus laevis. In 2001 they reported that mutant forms of frog Dishevelled, targeted to neural tissue, stopped neural tissue from elongating in grafts and explants, and that marker genes showed the tissue was still neural. In 2002 they asked where the signal mattered. Because the fate of each early frog cell is mapped, they could inject a disrupted Dishevelled into cells that would form the midline of the plate in some embryos, and into cells that would form its sides in others.

Midline targeting blocked convergent extension. The plate stayed wide, and the tube stayed open, even though the folds rose and moved inward. Lateral targeting, in the region where the folds form and fuse, left closure intact. Under those conditions the requirement for Dishevelled was in the midline, where narrowing sets the distance between the folds.

Mammals give separate evidence. The loop-tail mouse, whose homozygotes have a tube open from midbrain to lower spine, carries a mutation in Vangl2, identified in 2001. In 2007 the laboratory of Andrew Copp labelled the midline of these mouse embryos and found that it failed to extend before neurulation began, and closure never started.

Narrowing alone does not close a tube. Apical constriction at the hinges, pushing from the surface ectoderm, and adhesion where the fold tips meet all contribute, and their share differs by species and by level of the body. Mice lacking the actin-binding protein Shroom3, for example, fail to close the head folds.

”What happens if this doesn’t close?”

A neural tube defect is a failure of closure, and its name depends on where the failure is.

• Anencephaly. The head end fails to close. The exposed brain tissue degenerates, and most of the forebrain and the top of the skull are missing. Affected babies are stillborn or die soon after birth.
• “Spina Bifida”. The lower end fails, or the bone over it does. In spina bifida occulta one or more vertebral arches are unclosed but the cord lies in place under normal skin; it is often found by chance and is, strictly, a defect of bone over a closed tube. In a meningocele the meninges, the membranes around the cord, bulge through the gap in a sac while the cord stays in place. In a myelomeningocele the cord itself is in the open lesion. That form commonly brings weakness or paralysis below the lesion, loss of bladder and bowel control, and often hydrocephalus.
• Craniorachischisis. The tube is open from the midbrain down the spine, as if closure never started. In mouse terms, the first initiation site never fuses, as in loop-tail.

When does all this happen? Embryologists date human embryos by Carnegie stages, a series of 23 stages defined by structure, not by age, covering the first eight weeks after fertilization. The neural folds first meet at about stage 10. The head-end neuropore closes at about stage 11 and the tail-end neuropore at about stages 12 to 13, by the end of the fourth week after fertilization. Exact days differ between sources. Doctors usually count pregnancy from the last menstrual period, which adds about two weeks, so the same events fall at about six weeks of pregnancy by that count. Any prevention has to be in place before then.

Design problem

Two mouse mutants both have open neural tubes. In the first, the neural plate is short and wide from the start. In the second, the plate narrows normally, but the folds never meet. Say what failed in each, and propose one measurement and one experiment that would tell them apart.

The folic-acid trials: strong evidence, open mechanism

Folate is a B vitamin. Its reduced forms carry one-carbon units for making thymidylate and purines and for methylation reactions, as that lesson explains. Folic acid is the synthetic form used in tablets and fortified food.

The randomized evidence came from two trials. The Medical Research Council Vitamin Study, led by Nicholas Wald and published in 1991, enrolled women who had already had an affected pregnancy. Taken from before conception, folic acid at 4 milligrams a day cut the recurrence of neural tube defects in this high-risk group by about 72 percent; other vitamins showed no clear effect. In Hungary, Andrew Czeizel and István Dudás assigned women planning a pregnancy to a multivitamin with 0.8 milligrams of folic acid or to trace elements. Their 1992 report found six defects in the trace-element group and none with the multivitamin, evidence that a first occurrence can also be prevented.

Policy followed. In 1992 the US Public Health Service advised 0.4 milligrams of folic acid a day for women who could become pregnant. Since many pregnancies are unplanned, the United States and Canada began requiring folic acid in enriched grain products in 1998, and many countries have since done the same. Recorded rates of neural tube defects fell after fortification, though the size of the fall depends on how and where defects are counted.

Much remains unexplained. How folate prevents defects is not settled. Most mothers of affected babies are not folate deficient by ordinary tests, and some defects persist despite fortification. In mice, some mutants respond to folate and others do not. Other risk factors point to other routes: valproic acid, an anticonvulsant taken in early pregnancy, raises the risk of spina bifida, and so does diabetes in the mother before pregnancy. Rare variants in human planar-polarity genes have been reported in some affected people. Folic acid lowers the risk; it does not remove it.