Neural Tube Formation

What neural tube formation is

Every neuron in your brain, every axon in your spinal cord, every cell of your retina traces back to a flat oval of cells about the width of a few hundred micrometers, sitting on the back of a three-week-old embryo. That oval is the neural plate, and the act of rolling it into a sealed tube is neurulation. It is one of the most consequential origami problems in biology: a two-dimensional sheet has to bend, lift, meet itself along a seam, and fuse into a hollow cylinder — without leaks, kinks, or gaps — and it has to do so in a tightly bounded window of a few days.

The end product is the neural tube, a hollow epithelial cylinder running head to tail. Its anterior end balloons into the three primary brain vesicles (forebrain, midbrain, hindbrain); the rest becomes the spinal cord. The hollow lumen becomes the ventricular system of the brain and the central canal of the cord, filled later with cerebrospinal fluid. The cells lining the tube — the neuroepithelium — are the stem cells that will generate the roughly 86 billion neurons of the adult human brain. Get the folding right and you have a nervous system. Get it wrong by even a small margin and the consequences are catastrophic and often lethal.

How the sheet folds into a tube

Neurulation runs through four overlapping stages. (1) Formation: the dorsal midline ectoderm thickens from a squamous sheet into a tall columnar epithelium — the neural plate — under the influence of signals from the underlying notochord. (2) Shaping: the plate narrows and lengthens by convergent extension, cells intercalating between their neighbors so the tissue gets longer and skinnier (this is why the future spinal cord is a long groove, not a circle). (3) Bending: the plate flexes at discrete hinge points while the lateral edges, the neural folds, elevate toward the midline. (4) Closure: the apposed folds meet at the dorsal midline, the epithelia fuse, and the tube separates from the overlying surface ectoderm, which seals over the top as future skin.

The force that bends the sheet is apical constriction. Each neuroepithelial cell wears a contractile belt of filamentous actin and non-muscle myosin II just beneath its apical surface. When that belt contracts, the apex of the cell purses shut while the basal end stays broad — the cell becomes a wedge. Wedge a row of cells and the sheet kinks toward the apical side. The wedging is concentrated at the median hinge point (MHP), a midline furrow anchored over the notochord that forms the floor of the future tube, and at the paired dorsolateral hinge points (DLHPs), which curl the upper edges of the folds inward so they can meet. The molecular trigger is the actin-binding protein Shroom3, which localizes to the apical junctions and recruits Rho-kinase (ROCK), which phosphorylates myosin light chain and powers the contraction.

Closure is not a single seam zipping from one end to the other. In the mouse, closure starts at three discrete initiation sites and spreads bidirectionally; the human pattern most likely uses multiple sites too. The last regions to seal are the open ends — the anterior (rostral) neuropore, closing around day 25, and the posterior (caudal) neuropore, closing around day 28. The very tip of the tail forms by a separate route, secondary neurulation, in which a solid rod of tail-bud mesenchyme condenses and then hollows out (cavitates) into a lumen rather than rolling up from a sheet.

The molecular signals and gradients

Before the plate can fold it has to be told to exist. Neural induction is famously a "default" program: ectoderm left to itself, under the influence of BMP (bone morphogenetic protein) signaling, becomes epidermis. The embryonic organizer — Hensen's node in amniotes, the Spemann–Mangold organizer in amphibians — secretes BMP antagonists (Noggin, Chordin, Follistatin) that bind and neutralize BMP locally. Where BMP is silenced, the overlying ectoderm reverts to its default neural fate. This is the molecular cash-out of Hans Spemann and Hilde Mangold's 1924 transplantation experiment, which won Spemann the 1935 Nobel Prize.

Once the plate exists, opposing morphogen gradients pattern it along the dorsoventral axis. From below, the notochord and then the floor plate secrete Sonic hedgehog (Shh): high Shh near the floor specifies the floor plate and V3 interneurons, intermediate levels specify motor neurons, and the gradient reads out as distinct progenitor domains. From above, the dorsal epidermis and roof plate secrete BMP and Wnt, specifying sensory interneurons and the roof plate. The overlap of these counter-gradients gives every cell an address. A separate planar signal, the planar cell polarity (PCP) pathway (Vangl2, Celsr1, Dishevelled), orients cells within the plane of the sheet and drives the convergent-extension movements that narrow the plate; PCP mutations in mice produce craniorachischisis, the most severe neural tube defect in which the tube stays open along almost its entire length.

The players and conditions

• Neuroepithelial cells. The tall columnar cells of the plate. They both generate the folding force (apical constriction) and serve as the stem-cell pool for the whole CNS. Their nuclei migrate up and down with the cell cycle — interkinetic nuclear migration — giving the epithelium a deceptively layered look.
• Notochord. A transient midline rod of mesoderm. It is the source of the inductive and ventralizing Shh signal and physically anchors the median hinge point. In adults it survives only as the nucleus pulposus inside each intervertebral disc.
• Neural folds and neural crest. The crests of the rising folds are the birthplace of neural crest cells, which delaminate via epithelial-to-mesenchymal transition and migrate to build the peripheral nervous system, melanocytes, adrenal medulla, and most of the facial skeleton.
• Hinge points. One median (MHP) over the notochord, two dorsolateral (DLHPs). The DLHPs depend on BMP being switched off by Noggin at the dorsal tips; experimentally removing that inhibition prevents the folds from curling and the tube fails to close.
• Cytoskeletal machinery. Apical F-actin, non-muscle myosin II, Shroom3, ROCK, and the Rho GTPases. Disrupting actomyosin contractility — for instance with the myosin inhibitor blebbistatin — abolishes apical constriction and folding in explants.
• Folate and the one-carbon cycle. Rapidly dividing neuroepithelium needs folate-dependent one-carbon units to make thymidine and to methylate DNA. Folate deficiency or MTHFR variants impair this and raise neural tube defect risk — the single most important modifiable factor.

Human neurulation timeline

Day (post-fertilization)	Carnegie stage	Event
~16	CS7	Gastrulation; notochordal process forms beneath dorsal ectoderm
~18	CS8	Neural plate becomes morphologically distinct; neural groove appears
~20	CS9	Neural folds elevate; first somites; median hinge point forms
~22	CS10	Folds begin to fuse at multiple sites; neural crest delaminates
~25	CS11	Anterior (rostral) neuropore closes
~28	CS12	Posterior (caudal) neuropore closes; primary neurulation complete
~28–48	CS12–13	Secondary neurulation extends the caudal cord from the tail bud

The numbers

• Timing window. Primary neurulation spans roughly 10 days in humans (days 18–28). The two neuropores close just three days apart — a remarkably narrow target.
• Plate dimensions. The early human neural plate is on the order of a few hundred micrometers wide and lengthens to a few millimeters as it shapes by convergent extension.
• Cell wedging geometry. Apical constriction can shrink a cell's apical area severalfold while leaving the basal area broad; a relatively modest per-cell apical reduction, summed across thousands of cells in a hinge, produces the macroscopic ~180° fold.
• Shh gradient. Distinct progenitor domains read concentration thresholds spanning roughly an order of magnitude of Shh, integrated over time — the longer a cell sees high Shh, the more ventral its identity.
• Defect incidence. Neural tube defects occur at roughly 0.5–2 per 1000 births, varying by geography, diet, and folate-cycle genetics. Folic acid (400 µg/day periconceptionally) cuts the recurrence risk by about 70%.
• Stem-cell output. The neuroepithelium lining the tube ultimately yields the roughly 86 billion neurons and a comparable number of glia in the adult human brain.

Primary vs secondary neurulation

Property	Primary neurulation	Secondary neurulation
Starting material	Flat neural plate (epithelial sheet)	Solid rod of tail-bud mesenchyme
How the lumen forms	Sheet rolls up and folds enclose a space	Solid rod cavitates from inside out
Region built (human)	Brain through upper sacral spinal cord	Lowest sacral and coccygeal cord
Key mechanism	Apical constriction, hinge points, fold fusion	Mesenchymal condensation and cavitation
Timing (human)	~day 18–28	~day 28–48
Closure defect	Open spina bifida, anencephaly, craniorachischisis	Closed defects: tethered cord, lipomyelomeningocele
Boundary	The junction sits near the future second sacral vertebra (S2); the two modes overlap in a transition zone

Where it shows up — organisms, disease, and examples

• Spina bifida. Failure of the posterior neuropore to close around day 28. The spectrum runs from harmless spina bifida occulta (a bony gap with intact cord) to myelomeningocele, in which neural tissue herniates through the open vertebrae and is exposed — causing paralysis below the lesion and hydrocephalus. Fetal surgery to close the lesion in utero (the MOMS trial, 2011) improves outcomes over postnatal repair.
• Anencephaly. Failure of the anterior neuropore. The forebrain and overlying skull fail to form; the condition is uniformly fatal at or shortly after birth.
• Craniorachischisis. The most severe defect — the tube stays open from the midbrain to the low spine because the convergent-extension step never narrows the plate. In mice it is the signature phenotype of planar-cell-polarity mutants (Vangl2, "loop-tail").
• Folic acid fortification. After the 1991 MRC trial showed a ~70% reduction in recurrence, the United States mandated folic-acid fortification of enriched grain in 1998; population NTD rates fell sharply within a few years. Many countries now fortify flour.
• Valproate and diabetes. The anticonvulsant valproate raises NTD risk roughly tenfold by interfering with folate metabolism and histone deacetylases; poorly controlled maternal diabetes and maternal hyperthermia are also established risk factors.
• Neural crest and the face. Because crest cells are born at the closing tube, defects of crest emigration cause a distinct family of disorders — Hirschsprung disease (missing gut neurons), Waardenburg syndrome (pigment and hearing), Treacher Collins and DiGeorge syndromes (craniofacial).
• Model organisms. Xenopus (frog) and chick gave us the classic induction and hinge-point work; the mouse is the workhorse genetic model with over 200 genes known to cause NTDs when mutated; zebrafish form their cord partly by a "neural keel" rod that later cavitates, a useful contrast to amniote folding.

Common misconceptions and pitfalls

• The tube zips shut from one end like a single zipper. No — closure initiates at multiple discrete sites and spreads bidirectionally, with the neuropores and the junctions between waves sealing last. This multi-site geometry is exactly why defects cluster at predictable positions.
• Apical constriction alone explains folding. It is the headline force, but bending also needs basal expansion, cell-cycle-linked nuclear positioning, convergent extension to narrow the plate, and extrinsic pushing from the expanding surface ectoderm. Knock out only one and the others can partly compensate.
• Neural crest cells come from a separate germ layer. They originate from ectoderm at the neural-plate border, not a fourth germ layer — though their astonishing range of derivatives earns them the informal "fourth germ layer" nickname.
• Folic acid works by being a vitamin the embryo eats. Its benefit is mechanistic, not nutritional in the casual sense: folate supplies one-carbon units for thymidine synthesis and DNA methylation in the fast-dividing neuroepithelium. Taking it after the tube has closed (day 28) does nothing for NTDs — the window is periconceptional.
• The lumen is created when the folds meet. The future lumen is the apical space that already exists above the neuroepithelium; folding wraps the sheet around that space. In secondary neurulation there is no sheet at all — the lumen is hollowed out of a solid rod.
• The whole CNS forms by primary neurulation. The lowest spinal cord forms by secondary neurulation from the tail bud; this is why low sacral defects are often "closed" (skin-covered) rather than open.