Convergent Extension: How Cells Intercalate to Narrow and Lengthen an Embryo

What convergent extension is and where it happens

Convergent extension is a morphogenetic movement in which a tissue simultaneously narrows (converges) along one axis while lengthening (extends) along the orthogonal axis. Crucially, this happens with no net change in cell number or tissue volume — the shape change comes purely from cells rearranging their relative positions.

It is most dramatic during gastrulation and neurulation. In the frog Xenopus laevis, the dorsal mesoderm and the overlying neural plate converge toward the dorsal midline and extend the anterior-posterior (head-to-tail) axis. The notochord — the rod that defines the chordate body plan — is sculpted almost entirely by CE, refining from a broad field into a narrow rod one or two cells wide.

• Vertebrate mesoderm & neural plate — axis elongation and neural tube closure
• Notochord — in Xenopus, zebrafish, and the ascidian Ciona (whose notochord is exactly 40 cells)
• Drosophila germband — the classic invertebrate example, doubling body length in ~30 minutes

Because the axis it builds is fundamental to bilaterian body plans, CE is one of the most deeply conserved shape-changing behaviors in animal development.

The mechanism, step by step

Convergent extension proceeds through a stereotyped sequence that Ray Keller and colleagues termed mediolateral intercalation behavior (MIB):

• 1. Polarization. Initially cells are randomly protrusive. Under PCP cues, protrusive activity becomes restricted to the medial and lateral ends of each cell, making them bipolar and elongated along the mediolateral axis.
• 2. Traction. The medial and lateral lamellipodia grip neighboring cells and exert traction, so cells pull themselves between one another rather than crawling on a substrate.
• 3. Intercalation. Cells insert between their neighbors, so the number of cells across the width drops while the number along the length rises — the tissue converges and extends.
• 4. Junction remodeling (alternative/parallel route). In epithelia (Drosophila germband, mouse neural plate), cells instead shrink their anterior-posterior junctions to zero and grow new ones perpendicular — the T1 transition — and multicellular rosettes form and resolve.

Both routes are driven by polarized actomyosin: non-muscle myosin II contracts specific cell edges while actin-based protrusions extend others, converting molecular polarity into directional tissue flow.

Key molecules and characteristic numbers

The master regulator is the non-canonical Wnt / planar cell polarity (PCP) pathway — a β-catenin-independent branch of Wnt signaling. Its core components segregate to opposite cell edges to create planar polarity:

• Ligands: Wnt5a and Wnt11 (in zebrafish, wnt11/silberblick and wnt5/pipetail mutants fail to extend).
• Receptors/transmembrane core: Frizzled (Fz), Vangl (Van Gogh/Strabismus), and Celsr/Flamingo, an atypical cadherin.
• Cytoplasmic core: Dishevelled (Dvl) and Diego partner with Frizzled on one edge; Prickle partners with Vangl on the opposite edge.
• Effectors: the small GTPases RhoA and Rac1, ROCK kinase, and non-muscle myosin II, which generate the contractile force.

Quantitatively, a Xenopus dorsal-mesoderm explant can extend by a factor of ~2–4 over 1–3 hours. The Ciona notochord elongates from a wide plate to a stack of exactly 40 cells. Drosophila germband extension doubles the axis (~2.2×) in roughly 30 minutes, with individual junctions shrinking over minutes.

How it is studied, observed, and regulated

Convergent extension is famously tractable because it can be reproduced outside the embryo. The Keller explant (Keller sandwich) — a piece of dorsal marginal-zone tissue excised from a Xenopus gastrula — will converge and extend on its own on a coverslip, letting researchers film individual cells. Shih and Keller's 1992 papers used exactly this preparation to define MIB.

• Live imaging: confocal and light-sheet microscopy of fluorescently labeled membranes and myosin reveal bipolar protrusions, junction shrinkage, and rosettes in real time.
• Genetics: zebrafish PCP mutants (trilobite/vangl2, knypek/glypican-4, silberblick/wnt11) show short, broad axes; mouse PCP mutants (Vangl2 Lp, Dvl1/2/3, Celsr1 Crash) fail neural tube closure.
• Perturbation: dominant-negative Dishevelled or morpholino knockdown blocks extension without blocking cell fate.

Regulation is spatial and temporal: cells must polarize relative to the embryonic axis, so the PCP system reads global tissue cues (Wnt gradients, mechanical strain) and translates them into asymmetric protein localization and edge-specific myosin activity.

How it differs from related processes

Convergent extension is easy to confuse with other movements that also elongate tissues, but the mechanism is distinct:

• vs. oriented cell division: CE adds no cells; elongation via division depends on the mitotic spindle and proliferation. Real embryos often combine both.
• vs. cell-shape change: individual cells can stretch to elongate a tissue, but CE relies on cells changing neighbors, not merely stretching.
• vs. collective migration: in migration a whole cohort translocates across a substrate; in CE cells pull on each other and stay within the tissue.
• MIB (crawling) vs. junction remodeling: mesenchymal tissues use bipolar protrusive traction; epithelia use apical-junction shrinkage, T1 transitions, and rosettes — two solutions to the same geometric problem.

It is also distinct from, but closely linked to, gastrulation as a whole: gastrulation is the broad reorganization of germ layers, whereas convergent extension is one specific, PCP-driven cell behavior that operates within gastrulation and neurulation.

Significance, disease relevance, and open questions

Convergent extension is not just an embryological curiosity — its failure causes some of the most common and severe human birth defects. When the neural plate cannot converge and extend, the neural tube fails to close, producing neural tube defects such as spina bifida and craniorachischisis. Human mutations in PCP genes (VANGL1, VANGL2, CELSR1, PRICKLE1, SCRIB, FZD6) have been identified in patients with these defects.

• Kidney and skeleton: PCP-driven CE also shapes renal tubules; its disruption is linked to cystic kidney phenotypes and to shortened, disproportioned limbs (e.g., Robinow syndrome, involving WNT5A/ROR2).
• Model-organism power: Xenopus, zebrafish, and Drosophila remain the workhorses because CE is experimentally isolable in each.

Open questions remain vigorous: How exactly is planar polarity oriented relative to the whole-embryo axis — by chemical Wnt gradients, mechanical strain, or both? How do the crawling and junction-remodeling modes cooperate in the same tissue? And how are the forces integrated across thousands of cells so the tissue extends smoothly rather than buckling? These are active areas in developmental biology and biophysics today.

Two cellular mechanisms of convergent extension and how they compare to related processes

Process	Cell behavior	Force generator	Example tissue
CE by crawling (MIB)	Bipolar mediolateral protrusions exert traction on neighbors	Actin-rich lamellipodia + traction	Xenopus notochord / dorsal mesoderm
CE by junction remodeling	Shrinking of vertical (AP) junctions, T1/rosette transitions	Junctional non-muscle myosin II contraction	Drosophila germband, mouse neural plate
Convergent extension (net)	Cells swap neighbors; number constant	PCP-polarized actomyosin	Vertebrate body-axis elongation
Cell division-driven growth	Oriented mitoses add cells	Cell cycle / proliferation	Late axis growth, imaginal discs
Cell migration (collective)	Whole population translocates over substrate	Leading-edge protrusion + adhesion	Neural crest, border cells