Developmental Neuroscience · Unit 3: Coordinates: patterning the tube

Local Organizers: The Midbrain-Hindbrain Boundary

Hox genes give the hindbrain and spinal cord their addresses, but the front of the neural tube needs other instructions. This lesson follows one of them, the isthmic organizer at the midbrain-hindbrain boundary: how two repressors draw the line, how a small loop of genes keeps it alive, and how grafts, beads and mouse mutants showed the organizer to be both sufficient and necessary. Afterwards you can predict the brain of an embryo in which one part of the loop is missing, moved or placed somewhere new.

Watch first One Bead Builds a Midbrain · 2:37

Push a small bead soaked in one protein, FGF8, into the back of a chick embryo’s forebrain, and a second midbrain can grow there, built back to front.

That result, from 1996, is the end of a story this lesson follows. The question behind it is simple. By the time the neural tube has closed, its cells need to know where along the tube they sit, so that some build forebrain, some midbrain, some hindbrain. Who tells them?

Beyond the Hox border: the brain needs other organizers

In the lesson on Hox genes and the hindbrain you met one answer. Hox genes, switched on in a nested order along the tube, give each segment of the hindbrain and spinal cord an address. But Hox expression stops near the front of the hindbrain. The midbrain and forebrain have no Hox code at all. In the professor’s words: “Beyond the hox border: Need other organizers.”

An organizer, in embryology, is a group of cells that changes what its neighbours become. The first one found was Spemann’s organizer, the lip of the frog gastrula that can induce a whole second body axis, nervous system included. The centers in this lesson are smaller and later. A local organizer (also called a secondary organizer) is a small signaling center inside the already-formed neural tube that patterns the tissue around it. Think of it as a loudspeaker standing at a property line: it does not build the houses, but it tells the houses nearby what to build.

The best studied sits at the junction between midbrain and hindbrain, at a narrowing of the tube called the isthmus. In front of it, the midbrain gives rise to the tectum (in mammals, the colliculi). Just behind it, the first hindbrain segment, rhombomere 1, gives rise to the cerebellum.

Keep the professor’s three principles from the Hox lesson in view, because this organizer follows all three. Genes are switched on at specific places and times. Graded signals from opposing sources set up patterns. Those signals restrict transcription factors, which in turn control the genes of each region.

A line drawn by two repressors, then held by a loop

The boundary is drawn before the organizer does any organizing. Two transcription factors divide the front of the tube. Otx2 is on in the forebrain and midbrain. Gbx2 is on in the front of the hindbrain. Each represses the other, so a cell cannot keep both, and the edge between the two domains becomes sharp. In the chick, Millet and colleagues in Alvarado-Mallart’s group reported in 1996 that the back edge of Otx2 lies at this junction.

Is the line really set by where the two meet? Two mouse studies published side by side in 1999 moved it. Broccoli, Boncinelli and Wurst made Otx2 in territory that would become anterior hindbrain; the boundary and Fgf8 moved back and the cerebellum shrank. Millet and colleagues in Joyner’s laboratory made Gbx2 in the back of the midbrain; the boundary moved forward. The line follows the edge of each repressor.

At that line a set of genes switches on and keeps itself running:

  • Fgf8, a secreted growth factor, in a band just behind the boundary. It acts through FGF receptors, which are receptor tyrosine kinases (see that lesson), and through the Ras-MAPK relay (that lesson).
  • Wnt1, a secreted Wnt ligand, in a ring just in front of the boundary (see that lesson).
  • Engrailed-1 and -2 and Pax2 and Pax5, transcription factors on in a broad band across the boundary.

In the standard model each of these helps keep the others on, so the loop holds once started. Remove one and the rest fade.

The professor’s next phrase is “Cousin of Hox: engrailed is a key determinant of neural identity.” A cousin, not a sibling. Engrailed was named in the fly for a defect in body segments. Like the Hox genes, it encodes a homeodomain, a DNA-binding unit of about 60 amino acids that recognizes short DNA sequences. Unlike them, it does not sit in a Hox cluster and is not part of the Hox code. In the midbrain, Engrailed is graded, highest near the isthmus, and that gradient later tells the tectum which end is which, a topic for the lesson on topographic maps.

Sufficient: the graft and the bead

The professor’s logic for every experiment in this unit has two halves. A factor is sufficient if adding it where it is normally absent produces the structure. It is necessary if removing it takes the structure away. You need both.

The first sufficiency test used tissue, not molecules. Alvarado-Mallart’s group in Paris grafted pieces of young bird brain into new positions, often quail into chick, since quail cells can be told apart from chick cells under the microscope. In 1991 Martínez, Wassef and Alvarado-Mallart reported that metencephalic tissue from just behind the midbrain-hindbrain junction, grafted into the forebrain of a 2-day chick embryo, kept its own identity. The surprise was in the host. Forebrain cells next to the graft turned on Engrailed and later developed as midbrain. The junction could instruct its neighbours.

What was the message? Fgf8 was a candidate, since it is made right at the boundary. In 1996 Philip Crossley, Salvador Martínez and Gail Martin soaked small beads in FGF8 protein and placed them in the caudal forebrain, the diencephalon, of chick embryos. The paper reports that FGF8 protein had the same midbrain-inducing effect as isthmic tissue, and that in embryos allowed to develop further an extra midbrain formed.

The extra midbrain was built in reverse. The parts that normally lie nearest the isthmus formed nearest the bead, and parts that lie farther forward formed farther away, so the new midbrain was a mirror image of the normal one. That is what you expect if cells read their distance from a source. The bead also switched on genes normally expressed in the isthmus in the tissue around it, a sign that it had started the loop rather than only standing in for it.

Necessary: “What happens if you knock it out?”

In the mouse, a gene can be removed by gene targeting: the gene is altered in embryonic stem cells, which are then used to make mice carrying the change. The professor’s question for each gene in the loop is the same, and the answers line up.

  • Wnt1. In 1990 Andrew McMahon and Allan Bradley, and independently Kirk Thomas and Mario Capecchi, reported mice lacking Wnt1. They lack the midbrain and much of the cerebellum. In these mutants Engrailed turns on and is then lost. Making En1 from Wnt1’s own control elements rescues early development of the region, so a main job of Wnt1 here is to keep Engrailed on.
  • Engrailed. Mice lacking En1 lose most of the colliculi and cerebellum (Wurst, Auerbach and Joyner, 1994). Mice lacking En2 have a milder cerebellar defect. Mice lacking both lose more of the region than either single mutant.
  • Pax2 and Pax5. Losing both genes removes the posterior midbrain and cerebellum; each alone has a milder or strain-dependent effect.
  • Fgf8. Mice with no Fgf8 fail at gastrulation, so the brain question needs an allele that reduces the gene. Mice that make greatly reduced Fgf8, or lose it early from the region, lack midbrain and cerebellum.

In zebrafish, the mutant acerebellar carries a broken Fgf8 gene. Its boundary genes switch on and then fade, so in the fish Fgf8 is needed more to maintain the program than to start it.

Design problem

Using the logic of sufficiency and necessity, predict the brain in each case, and say which experiment tests necessity and which tests sufficiency. (a) A mouse that makes greatly reduced Fgf8. (b) A mouse in which Otx2 is forced on in the territory that would become anterior hindbrain. (c) A chick embryo with an FGF8 bead in the caudal forebrain.

One way to do it

(a) With too little Fgf8, the loop fails: Engrailed, Wnt1 and Pax2 are not maintained, and the midbrain and cerebellum are lost. This tests necessity.

(b) The boundary is drawn where Otx2 meets Gbx2. Forcing Otx2 backward pushes Gbx2 back, so the boundary and the Fgf8 band shift toward the tail, and less rhombomere 1 is left to make cerebellum, which shrinks. This tests sufficiency of the Otx2 edge to position the line: forcing Otx2 backward is a gain-of-function change that asks whether the line follows where the two repressors meet.

(c) The bead adds FGF8 where it is normally absent. The forebrain tissue around it turns on Engrailed and later forms a midbrain whose order is reversed, with isthmus-like parts next to the bead. This tests sufficiency.

Other organizers along the tube

The isthmus is one of several. At the very front of the neural plate, the anterior neural ridge expresses Fgf8; in mouse explants, removing it blocked the forebrain gene Foxg1 (also called BF1), and FGF8 protein could induce that gene. In the diencephalon, a thin stripe called the zona limitans intrathalamica makes Sonic hedgehog and, in chick experiments, patterns the thalamus on one side and the prethalamus on the other.

Not every organizer adds a signal. In zebrafish, Corinne Houart, Monte Westerfield and Stephen Wilson found in 1998 that a small strip of cells at the front edge of the neural plate, the anterior neural boundary, is needed for normal forebrain markers, and can induce them when moved. In 2002 Houart, Wilson and colleagues identified one of its products, Tlc, a secreted relative of the Wnt receptor Frizzled that dampens Wnt signaling nearby. In the fish, Wnt from behind pushes tissue toward a posterior fate, and Tlc lets the telencephalon form by holding it off. The same three principles apply: a place, a signal opposed by another, and transcription factors restricted by the result.

How we measure it

Quail-chick grafting

A piece of neural tube is cut from a quail embryo and placed into a chosen spot in a chick embryo of similar stage. Quail cells look different from chick cells under the microscope, so you can tell whether a new structure is made of donor cells or of host cells that the graft has instructed.

Protein-soaked bead implantation

A small bead is soaked in a purified protein and pushed into embryonic tissue, where it releases the protein locally. Control beads soaked in buffer alone show what the wound and the bead do by themselves. A positive result tests sufficiency.

Gene targeting in mice

A gene is altered in embryonic stem cells, which are then used to make mice carrying the change. A knockout removes the gene, a hypomorph reduces it, and a knock-in puts another coding sequence under its control. These test necessity and position.

In situ hybridization

A labelled RNA probe binds the messenger RNA of one gene in a fixed embryo, showing where that gene is on. Comparing the domains of Otx2, Gbx2, Fgf8 and Engrailed in normal and mutant embryos shows where a boundary lies and whether it has moved.

Finish line

Test yourself

8 questions

Practice and MCAT-style, with explained answers.

Next · Lesson 10

Ventral to Dorsal: Sonic Hedgehog and BMP in the Neural Tube

Graft a second notochord beside a chick neural tube, and the tube grows a second floor.

Continue →

← Lesson 8 · Head to Tail: Hox Genes, Retinoic Acid and Rhombomeres

FGF8b bound to an FGF receptor — the isthmic signal meeting its receptor
FGF8b bound to an FGF receptor — the isthmic signal meeting its receptor ● LIVE · RCSB PDB 2FDB
open on RCSB PDB ↗
Loading live structure 2FDB from RCSB PDB…

Find the small FGF8 ligand held between the two immunoglobulin-like domains of the receptor. The paper argued that the short stretch at FGF8b's front end, absent from the weaker FGF8a form, adds a contact with the receptor.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
FGF receptor 1 kinase domain — what the receptor turns on inside the cell
FGF receptor 1 kinase domain — what the receptor turns on inside the cell ● LIVE · RCSB PDB 1FGK
open on RCSB PDB ↗
Loading live structure 1FGK from RCSB PDB…

Look for the two lobes of the kinase with the ATP site between them. When FGF8 brings two receptors together, kinase domains like this phosphorylate each other and start the Ras-MAPK relay.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Engrailed homeodomain on DNA — the cousin of Hox at work
Engrailed homeodomain on DNA — the cousin of Hox at work ● LIVE · RCSB PDB 1HDD
open on RCSB PDB ↗
Loading live structure 1HDD from RCSB PDB…

One helix of the homeodomain lies in the major groove of the DNA and reads the bases, while the flexible arm at its start reaches into the minor groove. Hox proteins bind DNA in the same way.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Rosa-Magda Alvarado-Mallart

Rosa-Magda Alvarado-Mallart

With Salvador Martínez and Marion Wassef, grafted metencephalic tissue from just behind the chick midbrain-hindbrain junction into the forebrain of 2-day chick embryos and found that host tissue next to the graft turned on Engrailed and later took on midbrain character.

Alvarado-Mallart's laboratory in Paris worked with grafts in bird embryos, moving pieces of the young brain from one place to another and asking what each piece became and what it did to its new neighbours. Grafts between quail and chick are useful here because quail cells can be told apart from chick cells under the microscope, so donor and host can be followed separately. In 1991, with Salvador Martínez and Marion Wassef, she reported that tissue from the metencephalon, just behind the midbrain-hindbrain junction, grafted into the forebrain of a 2-day chick embryo, kept its own identity. The striking result was in the host: forebrain tissue beside the graft turned on the homeobox gene Engrailed and later developed as midbrain. In 1996 her group, with Antonio Simeone, mapped Otx2 in the chick and placed the edge of its domain at the same junction. The grafts made it possible to treat the boundary as a source of instructions and to go looking for the molecule that carried them.

✓ verbatim Martínez, Wassef & Alvarado-Mallart, Neuron 6:971 (1991); Millet, Bloch-Gallego, Simeone & Alvarado-Mallart, Development 122:3785 (1996).
Gail Martin

Gail Martin

With Philip Crossley and Salvador Martínez, showed that a bead soaked in FGF8 protein and placed in the chick caudal forebrain induces an extra midbrain with reversed polarity; her laboratory later made mouse Fgf8 alleles that reduce or remove the gene.

Martin's 1981 paper described pluripotent cell lines grown from normal mouse blastocysts and used the name embryonic stem cells; Martin Evans and Matthew Kaufman reported a parallel derivation the same year. Such cells later made targeted mouse mutants routine. At the University of California, San Francisco, her laboratory studied Fgf8, a gene for a secreted growth factor expressed in a narrow band just behind the midbrain-hindbrain boundary. The graft experiments had shown that this region could instruct its neighbours, and Fgf8 was a candidate for the message. In 1996 Philip Crossley, Salvador Martínez and Martin placed beads soaked in FGF8 protein in the caudal forebrain of chick embryos. Engrailed-2 turned on around the bead, and embryos allowed to develop went on to grow an extra midbrain whose parts lay in reverse order. Two years later her laboratory reported a series of mouse Fgf8 alleles, including one that makes much less of the gene's product, which let others ask what the boundary needs.

✓ verbatim Crossley, Martínez & Martin, Nature 380:66 (1996); Meyers, Lewandoski & Martin, Nature Genetics 18:136 (1998).
Andrew McMahon

Andrew McMahon

With Allan Bradley, made mice lacking Wnt-1 and found that they lack the midbrain and much of the cerebellum; Kirk Thomas and Mario Capecchi reported a Wnt-1 mutant independently the same year.

Wnt-1 began as int-1, a mouse gene that a virus could switch on to cause mammary tumours. In the developing mouse, McMahon and others found it expressed in the neural tube, including a ring just in front of the midbrain-hindbrain boundary. In 1990 McMahon and Allan Bradley used gene targeting in embryonic stem cells to make mice lacking the gene. The mutants lacked the midbrain and much of the cerebellum. In the same year Kirk Thomas and Mario Capecchi reported their own Wnt-1 mutant mice, with a similar brain defect. Later work from McMahon's laboratory found that in the mutants Engrailed turns on in the region and is then lost, and the tissue disappears. In 1996 Paul Danielian and McMahon made Engrailed-1 from the Wnt-1 gene's control elements in Wnt-1 mutants, and early development of the missing midbrain and anterior hindbrain was restored. That result put a signal and a transcription factor in order: Wnt-1 keeps Engrailed on, and Engrailed keeps the territory.

✓ verbatim McMahon & Bradley, Cell 62:1073 (1990); Thomas & Capecchi, Nature 346:847 (1990); Danielian & McMahon, Nature 383:332 (1996).
Alexandra Joyner

Alexandra Joyner

Studied the mouse Engrailed genes, finding that mice lacking En1 lose most of the colliculi and cerebellum, and with colleagues used gene swaps and misexpression to test how Engrailed and Gbx2 build the boundary.

Joyner's work on Engrailed began with finding mouse genes related to the fly gene engrailed, in work with Gail Martin's laboratory. Mice have two, En1 and En2, both expressed across the midbrain-hindbrain boundary. In 1994 Wolfgang Wurst, Bruce Auerbach and Joyner reported that mice lacking En1 die soon after birth and lack most of the colliculi and the cerebellum, while loss of En2 had earlier been found to cause a milder cerebellar defect. Were the two proteins doing different jobs, or the same job at different times? In 1995 her group replaced the En1 coding sequence with En2, and the brain defect of En1 mutants was rescued, so the difference lay mostly in when and where each gene is expressed. In 1999 her laboratory, with S. Millet and colleagues, reported that Gbx2 made in the posterior midbrain moves the boundary forward. Such experiments turned the boundary into something that could be moved by genetics.

✓ verbatim Wurst, Auerbach & Joyner, Development 120:2065 (1994); Hanks et al., Science 269:679 (1995); Millet et al., Nature 401:161 (1999).
Test yourself
Question 1 of 8 Score 0 / 0

Which description fits a local organizer such as the isthmus?
Sources & Integrity

Every structure, slide, simulation, and clip on this page is listed below with the source it came from. None of the 3D structures is drawn by hand or by AI: each one loads by accession ID from RCSB PDB or PubChem. This panel is generated from the lesson's own data file, so what it lists is what the page loads.

Molecular structures live data

  • FGF8b bound to an FGF receptor — the isthmic signal meeting its receptor — RCSB PDB 2FDB · coordinates CC0 · Human FGF8b with the ligand-binding region of FGFR2c, X-ray crystal structure (Olsen et al., Genes Dev., 2006)
  • FGF receptor 1 kinase domain — what the receptor turns on inside the cell — RCSB PDB 1FGK · coordinates CC0 · Human FGFR1 tyrosine kinase domain, X-ray crystal structure (Mohammadi et al., Cell, 1996)
  • Engrailed homeodomain on DNA — the cousin of Hox at work — RCSB PDB 1HDD · coordinates CC0 · Drosophila Engrailed homeodomain bound to DNA, X-ray crystal structure (Kissinger et al., Cell, 1990)

Scientist profiles & quotes

  • Rosa-Magda Alvarado-Mallart — Martínez, Wassef & Alvarado-Mallart, Neuron 6:971 (1991); Millet, Bloch-Gallego, Simeone & Alvarado-Mallart, Development 122:3785 (1996).
  • Gail Martin — Crossley, Martínez & Martin, Nature 380:66 (1996); Meyers, Lewandoski & Martin, Nature Genetics 18:136 (1998).
  • Andrew McMahon — McMahon & Bradley, Cell 62:1073 (1990); Thomas & Capecchi, Nature 346:847 (1990); Danielian & McMahon, Nature 383:332 (1996).
  • Alexandra Joyner — Wurst, Auerbach & Joyner, Development 120:2065 (1994); Hanks et al., Science 269:679 (1995); Millet et al., Nature 401:161 (1999).

Quotes are verbatim from the cited source. Profile text is original to this site and was checked against the sources cited.

Animated explainer

One Bead Builds a Midbrain · narration: an AI narrator voice (Gemini, Charon) reading narration written from this lesson's text · animation: AI-generated in code (Manim) by Claude, a schematic with no drawn molecular structures, plots computed from their equations · captions from the narration script.

Images & figures

This lesson carries no raster figures. Every structure is rendered live from RCSB PDB or PubChem, and every plot is drawn from its equation in the browser.

integrity Machines check the mechanics. The build confirms that every local file exists, that every question's answer key points at a real choice, and that any AI-generated scientific figure names the person who checked it; a weekly job re-resolves every accession ID against its database. Questions are original; MCAT-style items are written for this site and are not real AAMC questions. Banned sources (the Tymoczko textbook, publisher decks, copyrighted figures) are not used.

Like I'm five

Imagine a long row of houses where one family paints its fence red and the next family paints its fence blue. Right where the red and blue fences meet, a neighbour sets up a loudspeaker and tells the houses nearby what to build. In the young brain, that loudspeaker is a small group of cells. Houses close to it build one part of the brain, and houses farther away build another.
High school

The early brain is a tube. Cells in the front part make a protein called Otx2, and cells just behind make one called Gbx2. Each protein switches the other off, so the line between them is sharp. Right at that line, cells start making signals, including FGF8 and Wnt1, that tell nearby cells to become midbrain on one side and cerebellum on the other. Scientists showed this two ways. Adding the signal somewhere new, with a bead soaked in FGF8, made a second midbrain grow. Removing a gene for one of the signals in mice made the midbrain and cerebellum disappear.
College

The isthmic organizer is a local signaling center at the midbrain-hindbrain boundary. Its position is set by the interface between Otx2, expressed in forebrain and midbrain, and Gbx2, expressed in anterior hindbrain, which repress each other. Fgf8 is expressed just behind the interface and Wnt1 just in front of it, while the homeodomain genes En1 and En2 and the paired-box genes Pax2 and Pax5 span it; these genes depend on one another for maintenance. FGF8 acts through FGF receptor tyrosine kinases and the Ras-MAPK pathway. Isthmic grafts and FGF8 beads in the chick caudal forebrain induce Engrailed and ectopic midbrain with reversed polarity, showing sufficiency. Mouse mutants lacking Wnt1, En1, both En genes or Pax2 and Pax5, or with greatly reduced Fgf8, lose midbrain and cerebellum, showing necessity.
Graduate

Positioning and maintenance are separable. The Otx2-Gbx2 interface fixes where the isthmic program runs: forcing Otx2 into the En1 domain shifts the boundary and Fgf8 caudally and shrinks the cerebellum, while Gbx2 misexpressed in posterior midbrain shifts it rostrally. Once placed, the program is held up by cross-regulation. In Wnt1 mutants Engrailed starts and is then lost, and restoring En1 from Wnt1 control elements rescues early development of the region. In zebrafish acerebellar, an Fgf8 mutant, boundary markers start and then fade, and timed conditional deletion of Fgf8 in mouse suggests that structures nearer the isthmus need longer exposure to Fgf8. FGF8b, which binds FGF receptors more strongly than FGF8a, is the isoform linked to cerebellar induction in chick misexpression studies.

Open question: the FGF8 bead induces mirror-image midbrain, so the tissue reads something like distance from the source, yet the boundary also acts as a lineage restriction that cells of the two sides seldom cross. How the same signal sets graded identity on one side, a fate choice on the other and a cell-sorting border between them is still not fully explained.
A tumour gene in the midbrain

Wnt1 was first found as int-1, a site where mouse mammary tumour virus inserted into the genome and switched on a nearby gene, giving mammary tumours. Nothing in that origin pointed to the brain. When int-1 was knocked out in mice in 1990, the mutants lacked the midbrain and much of the cerebellum. The fly gene wingless turned out to be its relative, and in 1991 the family was renamed Wnt, joining the two names. A gene first studied as an oncogene had a main job in building one region of the brain.

— Nusse & Varmus, Cell 31:99 (1982); McMahon & Bradley, Cell 62:1073 (1990); Nusse et al., Cell 64:231 (1991).
Swap the gene, keep the brain

Mice lacking En1 lose most of the colliculi and cerebellum, while mice lacking En2 have a milder cerebellar defect. That could mean the two proteins do different things. Joyner's group tested it by placing the En2 coding sequence where En1's had been, so that En2 protein was made at En1's times and places. The brain defect of the En1 mutant was rescued. For this job the difference between the genes lay mainly in their control regions, not in the proteins they encode.

— Hanks, Wurst, Anson-Cartwright, Auerbach & Joyner, Science 269:679 (1995).
Protecting the forebrain from Wnt

In zebrafish, Corinne Houart, Monte Westerfield and Stephen Wilson removed a small strip of cells at the front edge of the neural plate during gastrulation, and forebrain markers failed to appear normally. Cells from that strip, moved farther back, induced forebrain genes in nearby tissue. In 2002 Houart, Wilson and colleagues identified one of its signals: Tlc, a secreted Frizzled-related protein that dampens Wnt signaling. This organizer worked mainly by quieting a signal, in the fish.

— Houart, Westerfield & Wilson, Nature 391:788 (1998); Houart et al., Neuron 35:255 (2002).

 ↑ Top