Developmental Biology: Chapter 6 — Morphogens, induction, and the main pathways

🔬 Key Illustrated Scientific Figure

Full-Screen Zoom

Click to enlarge on this page

6.1 Concentration, competence, and range #

6.1.1 The French flag #

A morphogen is a signal that forms a gradient from a local source and specifies more than one fate according to its concentration. Lewis Wolpert’s French flag is the diagram: cells read high, middle, and low as three different instructions, so one source and one decay can pattern a row. A signal that pushes every cell above one threshold into the same fate is an inducer, not yet a morphogen. The strict test is that at least two responses must be direct, at two concentrations, rather than a relay in which the first induced cell secretes a different signal.

Bicoid in the syncytial fly is a morphogen, with the caveat in Chapter 4 that it never has to cross a membrane. In a cellular embryo the classical demonstration is activin, a TGF-β-family ligand, on dissociated Xenopus animal-cap cells. Different concentrations turn on different genes — Brachyury at a lower dose, goosecoid at a higher one — in the cell that receives the ligand. Chapter 5 used that result for mesoderm. Sonic hedgehog in the ventral neural tube is the vertebrate case to remember, because one gradient is read into more than three fates.

The tube sees Shh from the notochord and, at still higher local levels, from the floor plate that the notochord induced. From ventral to dorsal the fates run: floor plate (FoxA2), V3 interneurons (Nkx2.2), motor neurons (Olig2), then V2, V1, and V0 interneurons. Dorsal interneurons are a different system, under BMP and Wnt from the roof plate. Neighbours cross-repress, so a shallow gradient becomes a sharp boundary: Nkx2.2 and Olig2 each try to extinguish the other, and the winner is the one whose signal is locally stronger. Gli proteins are the readers. Where Shh is high, Gli activators, chiefly Gli2, predominate. Where Shh is absent, Gli3 is processed to a transcriptional repressor. Target enhancers differ in affinity, so a low-affinity site fills only at high Gli activity.

Real flags are maintained, not painted once. Hedgehog upregulates its own receptor Patched, which binds and sequesters the ligand, steepening the gradient and buffering it against small changes in production. Opposing gradients are the usual geometry: Shh from the floor and BMPs from the roof, or FGF from the tail bud and retinoic acid from the somites. Time matters as well as level. In the limb, how long a cell sees Shh changes its digit, which a single snapshot of concentration would miss. Diffusion over a field of many cell diameters is also not a free walk through water. Lipid-modified ligands stick, heparan sulphate binds them, and some signals travel on cytonemes, thin projections that deliver ligand to a chosen target. Biophysics can calculate a length constant. These notes only require the biological consequence: a source, a sink, and a reader that thresholds.

6.1.2 Competence and sequential induction #

Competence is the ability to respond. It depends on having the receptor, the transduction machinery, and chromatin that the pathway can open, and it is temporary. Spemann found that early gastrula ectoderm responds to an organiser by becoming neural, and that older ectoderm does not. The signal can be unchanged while the audience has left. An experiment that ignores the age of the responding tissue is not a test of the signal.

Induction is instructive when the signal chooses the fate, and permissive when the tissue is already specified and the signal merely allows that programme to run — a matrix, a survival factor, a blood supply. The apical ectodermal ridge of the limb, in the next chapter, is closer to permissive maintenance of outgrowth than to a list of different instructions for upper arm, forearm, and hand. The notochord’s induction of floor plate is instructive.

Sequential induction is a chain: tissue A induces B, B induces C, and often C signals back. The classical chain is the lens. The optic vesicle, an outgrowth of the forebrain, contacts head surface ectoderm and induces a lens placode. The lens then helps the vesicle become an optic cup and induces the overlying ectoderm to start a cornea. Spemann read this as a single instructor, the optic vesicle, acting on naive skin.

That account is incomplete, not empty. Work from Grainger and others showed that a lens-forming bias is already spread through a region of head ectoderm before the vesicle arrives, placed there by earlier signals from anterior neural tissue and mesoderm. The vesicle cannot induce a lens from flank ectoderm that never received that bias. Competence was restricted before the “inducer” appeared, and the inducer is one necessary step in a distributed sequence. Chapter 7 builds the eye on that correction. The habit to take from it is general: a famous grafting result names a sufficient source at one stage. It does not prove that source was the first, or the only, instructor.

6.1.3 Paracrine and juxtacrine #

Paracrine signals are secreted and act on nearby cells. Morphogens are paracrine, even when their travel is helped by cytonemes or matrix. Juxtacrine signals require contact. The ligand is in the membrane of the sending cell, and the receptor is in the membrane of the neighbour. Notch is the juxtacrine system in this course. Autocrine signalling acts on the producer. An intact animal cap becoming epidermis, because it bathes itself in BMP, is autocrine. Endocrine signalling uses the circulation. Retinoic acid and a few other small molecules sit awkwardly between paracrine and endocrine: they are made locally, they diffuse, and they are not proteins. The five pathways below are the ones reused until the names become a grammar. Learn the receptor class and what arrives at the DNA, not a list of synonyms.

6.2 Five pathways #

6.2.1 Hedgehog #

Hedgehog protein is covalently modified by cholesterol and by palmitate, so it does not float off the producing cell unaided. Dispatched helps it leave. In vertebrates the receptor machinery is parked in the primary cilium, a microtubule antenna that Cell Biology should not be confused with the motile cilia of the node. Those motile cilia break left–right symmetry. The primary cilium is a signalling organelle, and it does not need to beat.

Without ligand, Patched sits in the ciliary membrane and keeps Smoothened out. Gli proteins, especially Gli3, are phosphorylated by PKA, GSK3, and CK1 and cleaved to a repressor, so target genes stay off. With ligand, Hedgehog binds Patched, Patched leaves the cilium, and Smoothened enters. Cleavage of Gli stops. Gli2 in its activator form reaches the nucleus and transcription starts. Patched and Gli1 are themselves targets, so the pathway reports that it has fired and, via Patched, limits how far ligand spreads. Drosophila uses the same Patched–Smoothened–Ci logic but routes it through the plasma membrane of cells that have no primary cilium. A ciliary mutation in a vertebrate therefore looks, in part, like a Hedgehog mutation. A fly does not have that particular failure mode.

The neural-tube gradient above is the demonstration that concentration is read as many fates. The limb uses the same pathway as a polarising centre, and the midline uses it to split the eye field. One transducer, three developmental sentences. Cyclopamine, the teratogen from corn lily, binds Smoothened and blocks the transducer, which is why it appears again when cyclopia is discussed.

6.2.2 Wnt #

Canonical Wnt signalling is a decision about the stability of β-catenin. Without Wnt, a destruction complex — Axin, APC, GSK3, and CK1 — phosphorylates β-catenin, which is ubiquitinated and degraded. TCF/LEF sits on target genes with the repressor Groucho and keeps them off. Wnt binds Frizzled and the co-receptor LRP5/6, Dishevelled is recruited, and the destruction complex is inhibited at the membrane. β-catenin accumulates, enters the nucleus, displaces Groucho, and activates transcription with TCF. The early frog uses this pathway dorsally, before there is even much zygotic transcription, because cortical rotation locally disabled the destruction complex. Wingless in the fly segment is the same pathway after cellularisation, acting across one cell row.

Non-canonical Wnt signalling does not stabilise β-catenin. The planar-polarity branch, through Frizzled and Dishevelled to Rho and Rac, orients the cytoskeleton. Convergent extension of the frog axis and of the neural plate is this branch. A Wnt11 that participates in early dorsal β-catenin signalling and a later Wnt5a/Wnt11 deployment that drives intercalation are not one switch labelled “Wnt on”. Chapter 5’s warning about the two phases of Wnt is a warning about which branch, and when.

6.2.3 TGF-β and BMP #

Ligands of this family are dimers: Nodal, activin, Vg1, the BMPs, and others. The receptors are transmembrane serine/threonine kinases. Ligand brings a type II receptor, which is constitutively active, together with a type I receptor. Type II phosphorylates type I. Type I phosphorylates a receptor-regulated Smad. Here the family splits, and the split is why “TGF-β” must not be used as a synonym for “ventral” or “dorsal”.

The BMP branch phosphorylates Smad1, Smad5, or Smad8. The activin/Nodal/TGF-β branch phosphorylates Smad2 or Smad3. Either R-Smad must pair with Smad4 to regulate transcription in the nucleus. Inhibitory Smads, Smad6 and Smad7, and extracellular antagonists close the pathway. Chordin, noggin, and follistatin antagonise BMPs. Lefty and Cerberus antagonise Nodal, and Cerberus also antagonises BMP and Wnt. High Nodal/Smad2 specifies organiser mesoderm. High BMP/Smad1 specifies epidermis and ventral mesoderm. Same receptor class, different ligand, different Smad, opposite territory. FGF connects the branches because MAPK phosphorylation of the Smad1 linker inhibits the BMP Smads. A neural plate is low BMP signalling, not an absence of TGF-β-family biochemistry.

6.2.4 FGF #

FGF ligands bind receptor tyrosine kinases. Heparan sulphate proteoglycans are co-receptors, not decoration, and without them the signal is poor. Dimerisation and autophosphorylation of the FGFR recruit the Ras–MAPK cascade. ERK then phosphorylates transcription factors, Ets-family proteins among them, and also the Smad1 linker just mentioned. PI3K and phospholipase Cγ are alternative branches; the developmental events in these notes run mainly through MAPK. The intracellular order Grb2–Sos–Ras–Raf–MEK–ERK is Cell Biology. Re-deriving it here would bury the embryology.

What the embryo uses FGF for is outgrowth and a posteriorising or maintenance signal. FGF8 from the apical ectodermal ridge keeps limb mesenchyme proliferating. FGF8 in the tail bud is the posterior high end of the somite wavefront. Mesoderm induced by Nodal is not stably maintained if FGF signalling is blocked. None of these is a different receptor class. They are the same receptor tyrosine kinase aimed at different enhancers in competent cells.

6.2.5 Notch and lateral inhibition #

Notch is juxtacrine. Delta, or Serrate/Jagged, is the ligand in the sending membrane. Notch is the receptor in the receiving membrane. Productive binding requires the sending cell to endocytose its ligand, which pulls on Notch. An ADAM protease then sheds the Notch ectodomain, and γ-secretase cuts inside the transmembrane region, releasing the Notch intracellular domain, NICD. NICD goes to the nucleus, binds CSL (Suppressor of Hairless in the fly, RBPJ in vertebrates), recruits Mastermind, and turns CSL from a repressor into an activator. The characteristic targets are Hes genes, encoding repressors that shut off the sending cell’s programme in the receiver.

Lateral inhibition uses that link to space two fates. In the fly epidermis, proneural genes of the achaete-scute complex promote both a bristle fate and Delta expression. A cell that is slightly ahead signals harder, forces its neighbours’ Notch pathway on, and those neighbours express Hes proteins that repress achaete-scute and Delta. The leader becomes the sensory mother cell. The neighbours are diverted to epidermis. The same logic stops every cell of the vertebrate neural plate from becoming a neuron: neurogenins play the proneural role, Delta and Notch the inhibitory one. A small bias is amplified into a salt-and-pepper pattern. No long-range gradient is required, and a French flag would be the wrong drawing.

Notch has other modes. Fringe, a glycosyltransferase, modifies Notch so that it prefers one ligand over another, and that bias helps boundaries: the dorsoventral wing margin in the fly, and the somite boundaries already met as Lunatic fringe in the clock. Notch also keeps some stem-cell populations from differentiating. Those are real uses. Lateral inhibition is the one that explains a spaced pattern, and it should not be redescribed as a morphogen gradient with the name changed.

6.3 Key terms, summary, and self-check #

Key terms. Morphogen; French flag; threshold; competence; instructive induction; permissive induction; sequential induction; paracrine; juxtacrine; Patched; Smoothened; Gli; Frizzled; β-catenin; Smad; FGFR; NICD; lateral inhibition.

Summary. A morphogen specifies more than one fate by concentration, and thresholds are sharpened by enhancer affinity, cross-repression, and feedback on the ligand. Competence decides whether a tissue can hear the signal at that stage, which is why the optic vesicle is not the sole author of the lens. Hedgehog switches Gli from repressor to activator via Patched and Smoothened, in vertebrates inside the primary cilium. Canonical Wnt stabilises β-catenin; planar-polarity Wnt orients intercalation instead. TGF-β-family receptors are serine/threonine kinases split into a Smad1 BMP branch and a Smad2 Nodal branch. FGF is a receptor tyrosine kinase read mainly through MAPK. Notch is contact-dependent cleavage of NICD, and lateral inhibition uses it to choose one cell and silence its neighbours.

Self-check.

1. Why is a signal that induces only one fate, at one threshold, not a morphogen? Name one vertebrate gradient that specifies more than three.

2. What did the optic-vesicle graft get right about the lens, and what earlier bias did it miss?

3. For Hedgehog and for BMP, name the receptor class and the protein family that binds DNA at the end of the pathway.

4. Which Wnt pathway stabilises β-catenin, and which one drives convergent extension?

5. Why does lateral inhibition space bristles, where a French-flag gradient would not be the right mechanism?

—

—

—