Neurulation and Neural Tube Formation

Neurulation is the sequence of morphogenetic movements by which the neural plate — a thickened, elongated region of dorsal ectoderm — rolls into the neural tube. The neural plate first appears after gastrulation, when the underlying notochord and paraxial mesoderm signal to the overlying ectoderm, biasing it toward neural rather than epidermal fate (a process called neural induction, covered in depth elsewhere). The induced ectoderm thickens into a columnar pseudostratified epithelium — the neural plate — flanked on either side by the neural folds, ridges of tissue that will eventually meet at the dorsal midline. Neurulation itself unfolds in a fixed sequence: the neural plate narrows and elongates (a movement called convergent extension), its cells constrict at their apical (inner) surface to form a midline groove, the neural folds rise and arch over that groove, and finally the folds meet and fuse at the dorsal midline, sealing off a hollow tube that pinches away from the overlying, now-continuous surface ectoderm (the future epidermis). Cells at the very tips of the folds — the neural crest — delaminate from the closing tube's dorsal margin and migrate throughout the embryo to form much of the peripheral nervous system, pigment cells, and craniofacial skeleton. Vertebrates achieve this fold by two related but distinguishable strategies. In primary neurulation, seen along most of the future brain and upper-to-mid spinal cord, the flat plate bends and its edges fuse directly. In secondary neurulation, which builds the lowermost spinal cord (the tail-bud region), a solid cord of mesenchymal cells condenses and then cavitates — hollows out from the inside — to form the tube's lumen, without ever passing through a flat, folded intermediate. Both processes converge on the same final architecture: a fluid-filled tube whose wall is a pseudostratified neuroepithelium, which will go on to generate essentially every neuron and glial cell of the CNS through the process of neurogenesis.

Step 1 — Induction and plate formation. Following gastrulation, the notochord (axial mesoderm) and adjacent tissues secrete signaling molecules — including Noggin, Chordin, and Follistatin — that antagonize BMP4 (Bone Morphogenetic Protein 4) signaling in the overlying ectoderm. Because default, unopposed BMP signaling drives ectoderm toward epidermal fate, local suppression of BMP is what permits a dorsal strip of ectoderm to adopt neural identity instead. The result is the neural plate: a spoon-shaped sheet, broad anteriorly (future brain) and narrow posteriorly (future spinal cord), continuous at its lateral edges with the neural folds.

Step 2 — Shaping the plate: convergent extension. Before it folds, the plate must narrow mediolaterally and lengthen along the anterior-posterior axis. Cells intercalate — cells from either side of the midline insert themselves between one another — a movement powered by the planar cell polarity (PCP) pathway, which orients cytoskeletal machinery within the plane of the epithelium (using core PCP proteins such as Vangl2, Celsr1, and Frizzled, homologous to pathways first defined in Drosophila wing patterning). This convergent extension elongates and narrows the plate, a prerequisite for correct closure — mice with mutations in PCP genes (e.g., loop-tail, a Vangl2 mutant) exhibit severe neural tube closure failure (craniorachischisis, in which the tube fails to close along its entire length).

Step 3 — Bending the plate: apical constriction. Within the neuroepithelium, individual cells actively change shape. Actin filaments and non-muscle myosin II assemble into a contractile ring at the apical (lumen-facing) surface of each cell. Contraction of this actomyosin ring shrinks the apical surface while the basal surface remains wide, forcing each cell into a wedge shape. Because all the cells across the plate constrict simultaneously, the sheet bends toward its apical (inner, future ventricular) surface — exactly as a flat piece of paper curls when you compress one face of it. Constriction is concentrated at defined hinge points:

• The median hinge point (MHP), at the ventral midline, overlying and anchored to the notochord, which itself secretes Sonic hedgehog (Shh) promoting hinge-point formation.
• Paired dorsolateral hinge points (DLHPs), located more laterally, which appear at specific axial levels and are essential for closure of the hindbrain and lower spinal regions.

Step 4 — Elevation and fusion of the neural folds. As hinge points bend the floor of the plate downward and the folds arch up and inward, the tips of the neural folds — carrying neural crest cells — approach each other at the dorsal midline. Cell-surface adhesion molecules, including N-cadherin and various cadherin-switching events, mediate the final fusion: epithelial cells from the two folds interdigitate and seal, converting the open groove into a closed tube with a continuous roof plate. Immediately dorsal to the fused neural tube, the surface ectoderm also re-fuses into a continuous sheet, re-covering the embryo and physically separating the neural tube from the skin.

Step 5 — Multi-site, zippering closure (mammals). In mice and humans, the tube does not close from one end to the other in a single continuous zip. Instead, closure initiates independently at several discrete points along the body axis and then proceeds bidirectionally from each, like several zippers being pulled outward from separate starting points until they meet and fuse the gaps between them. In the mouse, three closure initiation sites are well described; in humans, the classical model similarly proposes multiple closure sites, though the precise number and locations remain an area of ongoing anatomical debate. The last regions to close are the rostral anterior neuropore (closing around day 25–26 in humans) and the caudal posterior neuropore (closing around day 27–28). Failure at these specific, last-to-close points explains why neural tube defects cluster at cranial (anencephaly) and lumbosacral (spina bifida) sites rather than occurring randomly along the body.

Step 6 — Detachment and lumen stabilization. Once fused, the neural tube separates from the overlying surface ectoderm and sinks beneath it, its lumen becoming continuous and fluid-filled — the precursor of the future ventricular system and central canal, later filled with cerebrospinal fluid. The neuroepithelial cells lining this lumen retain apical-basal polarity and proliferate rapidly, undergoing interkinetic nuclear migration (nuclei moving up and down the cell as they progress through the cell cycle) to build the thickened neural tube wall that will generate essentially all CNS neurons and macroglia.

The idea that neurulation depends on instructive signals from a specific tissue, rather than being an autonomous property of ectoderm, was established by the single most famous experiment in developmental biology: the Spemann-Mangold organizer graft (1924). Hans Spemann and his doctoral student Hilde Mangold worked with two newt species distinguishable by natural pigmentation — the lightly pigmented Triturus cristatus and the darkly pigmented Triturus taeniatus. Mangold microsurgically excised the dorsal blastopore lip (the "organizer" region) from an early gastrula of one species and grafted it into the ventral, belly-side ectoderm of a host embryo of the other species — tissue that would normally have become epidermis, not neural tissue. The result was extraordinary: the graft induced formation of a second, nearly complete embryonic axis, including a second neural tube, on the host's belly, and pigment tracing showed this secondary neural tube was made mostly of host tissue — the graft had induced surrounding ectoderm to adopt neural fate rather than simply forming a tube from its own cells. This proved that the dorsal lip actively induces neural identity in adjacent ectoderm; neural tissue is not a default fate that appears anywhere ectoderm happens to sit, but a specific instructed outcome of intercellular signaling. The work won Spemann the Nobel Prize in Physiology or Medicine in 1935 (Mangold died in a household accident in 1924, before the prize, and so was never nominated). For decades the molecular identity of the "organizer signal" remained unknown — this was famously called the "Spemann organizer problem." It was resolved in the early 1990s through work in Xenopus laevis by Richard Harland, Eddy De Robertis, and colleagues, who showed that injecting mRNA encoding the secreted proteins Noggin and Chordin into ventral ectoderm was sufficient to induce ectopic neural tissue and secondary axes, phenocopying Spemann-Mangold's graft. Complementary loss-of-function work showed that ectopic BMP4 expression suppresses neural fate and promotes epidermis, and that Noggin/Chordin work by directly binding and sequestering BMP4, preventing it from activating its receptor. This established the modern molecular model: neural fate is the default state of ectoderm once BMP signaling is blocked — the "default model of neural induction" — with the organizer's role being to secrete BMP antagonists rather than a novel "neuralizing" positive signal per se (a simplification, since additional signals such as FGF also contribute in some species and depend on precise timing). A separate, mechanically focused line of evidence came from direct cell biological and genetic dissection of the folding process itself. Laser ablation and drug studies (e.g., using the myosin inhibitor blebbistatin in chick and Xenopus neural plates) showed that disrupting apical actomyosin contraction specifically blocks hinge-point bending without blocking convergent extension, demonstrating these are mechanistically separable processes. Mouse genetics provided convergent confirmation: mutations in planar cell polarity genes (Vangl2, Celsr1, Scrib) cause craniorachischisis — complete failure of neural tube closure along the entire body axis — because convergent extension fails and the plate never narrows enough to allow the folds to meet, while mutations affecting hinge-point-specific signaling (e.g., disruptions of Shh signaling from the notochord) cause more localized closure defects. Together, these genetic and pharmacological experiments in model organisms (mouse, chick, Xenopus, zebrafish) established the modular, hierarchical logic described in the Mechanism section: induction sets the pattern, planar cell polarity shapes the plate, and apical constriction at defined hinge points bends it.

Example 1 — Reading a mouse genetic cross for neural tube defect risk. Suppose a heterozygous Vangl2^Lp/+ mouse (the well-studied loop-tail mutant, a recessive planar-cell-polarity allele) is crossed to another heterozygous Vangl2^Lp/+ mouse. Homozygous Lp/Lp embryos develop craniorachischisis (fully open neural tube) and typically die perinatally; heterozygotes are viable but sometimes show milder tail-kink phenotypes reflecting mild PCP disruption; wild-type (+/+) is unaffected. A standard Mendelian monohybrid cross gives:

So the expected ratio is 1 : 2 : 1, but only 25% of embryos show the severe phenotype. Now suppose an investigator observes, across many litters, only 20 severely affected pups out of 100 total offspring, instead of the expected 25 (i.e., 25/100). This deficit is expected and biologically meaningful: because craniorachischisis is a lethal, severe morphological defect, some Lp/Lp embryos die and are resorbed in utero before birth, so the observed live-birth ratio underrepresents the true Mendelian frequency at conception. This distinction — genotypic ratio at fertilization versus phenotypic ratio at the observation point (birth) — is a classic trap when interpreting neural tube defect genetics in mice, and it parallels why human population studies of neural tube defects must include stillbirths and terminated pregnancies, not just live births, to estimate a true incidence.

Example 2 — A clinical/epidemiological calculation: folate and neural tube defect risk. Neural tube defects (anencephaly, spina bifida, encephalocele) occur in roughly 1 in 1,000 pregnancies in populations without folate fortification, though rates vary substantially by region (historically higher in parts of the UK and China, lower in parts of Scandinavia). Randomized and observational studies from the 1990s (notably the UK Medical Research Council Vitamin Study, 1991) showed that periconceptional folic acid supplementation (4 mg/day in high-risk women with a prior affected pregnancy) reduced recurrence risk by about 72%, and lower-dose supplementation (0.4 mg/day) in general-population studies reduces first-occurrence risk by roughly 50–70%. Suppose a country with a baseline neural tube defect rate of 1.2 per 1,000 births introduces mandatory folic acid fortification of flour, achieving a measured 45% reduction in population incidence (a realistic figure from several national fortification programs, e.g., the U.S. and Canada in the late 1990s). We can calculate the new expected rate: New rate = baseline rate × (1 − reduction) = 1.2 per 1,000 × (1 − 0.45) = 1.2 × 0.55 = 0.66 per 1,000 births For a country with 4,000,000 births per year, this converts an expected 4,800 affected pregnancies annually (at baseline) into roughly 2,640 — a reduction of over 2,000 cases per year. This link — folate is required for one-carbon metabolism supporting the rapid DNA synthesis and methylation needed by the fast-dividing neuroepithelium during the fourth week of gestation — is why folate fortification is now recommended by the WHO and mandated in over 80 countries, making neurulation biology one of the most direct translations of embryology into preventive public health policy.

Hans Spemann trained as a zoologist in Germany and became the foremost experimental embryologist of the early twentieth century, developing microsurgical techniques — including fine glass needles and hair-loop ligatures — to manipulate amphibian embryos at a scale and precision no one had previously achieved. His central question was whether different regions of an early embryo were already committed to specific fates, or whether their fate depended on signals from their surroundings. Working with newt embryos at the University of Freiburg, he and his students performed transplantation experiments across gastrula stages, discovering that some regions retained the flexibility to change fate depending on their new location (were not yet determined), while a specific region — the dorsal blastopore lip — behaved completely differently: wherever it was placed, it organized a new, coordinated embryonic axis around itself. Hilde Mangold, Spemann's graduate student, performed the decisive transplantation herself as part of her doctoral thesis, meticulously using differently pigmented donor and host species so that induced tissue could be traced by cell origin under the microscope — a controls-based design remarkably rigorous for its time. She died in 1924, at age 26, in a gas explosion in her home, before the full impact of the work she had carried out was recognized; the landmark paper was published under both their names that same year.

Understanding neurulation directly shapes prenatal public health policy: mandatory folic acid fortification of staple grains, now implemented in more than 80 countries, is grounded in the mechanistic link between one-carbon metabolism and neuroepithelial proliferation during closure, and it has reduced neural tube defect incidence by roughly half in fortified populations. Prenatal screening — maternal serum alpha-fetoprotein testing and detailed ultrasound around 18–20 weeks — exploits the anatomical consequence of failed closure (an open neural tube leaks alpha-fetoprotein into amniotic fluid) to detect defects before birth. Fetal surgery for spina bifida, pioneered and validated by the randomized MOMS trial (Management of Myelomeningocele Study, published 2011), directly applies knowledge of tube anatomy: surgeons close an open lumbosacral lesion in utero, before birth, significantly improving outcomes such as hindbrain herniation and the need for postnatal shunting, compared with postnatal repair. In teratology and drug safety, knowledge of the precise closure window (roughly days 21–28 in humans) explains why certain drugs — most notably the anticonvulsant valproic acid — carry black-box warnings for pregnancy, since exposure during this narrow interval interferes with hinge-point bending and closure, elevating neural tube defect risk several-fold above baseline. In regenerative and stem cell biology, the molecular logic of neural induction (BMP inhibition driving neural fate) is used directly in the laboratory: protocols for generating neurons and cerebral organoids from human induced pluripotent stem cells rely on "dual SMAD inhibition" — simultaneously blocking BMP and Activin/Nodal signaling — a modern, in-a-dish recapitulation of what Noggin and Chordin do in the embryo. In evolutionary and comparative biology, the distinction between primary and secondary neurulation, and the diversity of closure-site number and location across vertebrates (from fish to birds to mammals), informs how the vertebrate body plan itself evolved, linking neurulation to the study of Neural Induction and Morphogen Gradients that pattern the tube once it is closed, and ultimately to the Axon Growth Cones and Guidance Cues that wire the neurons this tube goes on to generate.

Ask the tutor: How do dorsoventral morphogen gradients like Sonic hedgehog and BMP pattern the neural tube after it closes?