The Neurotrophic Hypothesis Tested

Inject a newborn mouse with an antiserum against one protein, and its sympathetic ganglia waste away. Give it extra of the same protein, and they swell.

The protein is nerve growth factor, NGF. Results like these tempt a strong conclusion: NGF decides how many sympathetic neurons an animal has. The neurotrophic hypothesis makes a narrower claim than that, and the tests of it show where the line falls. A survival factor can decide how many of the neurons that were made stay alive. It cannot add neurons that were never made.

Targets govern PNS development

You met the discovery in that lesson: a mouse tumour grafted into a chick embryo drew in sensory and sympathetic fibres, an activity it released could do the same in a culture dish, and Stanley Cohen purified that activity as a protein. The question here is what the protein does in a normal animal.

The groundwork was laid in chick embryos. A target is the tissue an axon grows to and connects with, such as a limb muscle or a gland. In 1934 Viktor Hamburger removed one wing bud from young chick embryos and found the neurons serving that limb reduced on the operated side. A small ganglion at the end of an experiment can mean two things, though. Fewer neurons might have been made, or the normal number might have been made and then lost. In 1949 Hamburger and Rita Levi-Montalcini followed chick spinal ganglia stage by stage and found that many neurons formed and grew fibres before the deficit appeared, with degenerating cells conspicuous later.

Chris Deppmann draws three lessons from this early work:

1. Neurons are overproduced.
2. A target-derived factor exists.
3. There is an implicit idea of competition.

The third is implicit because nobody had yet shown that neurons fight over the factor. That is what the hypothesis would have to earn. How many neurons die, and when, is the subject of that lesson.

Necessary and sufficient: adding NGF, removing it, deleting it

Two words do most of the work in this lesson. A factor is sufficient if it is enough, by itself, to produce an effect. It is necessary if the effect does not happen without it. Each experiment below tests one of these and no more.

Sufficiency. In 1960 Levi-Montalcini and Barbara Booker gave the purified salivary-gland protein to young mammals and reported greatly enlarged sympathetic ganglia. Purified factor injected into chick embryos likewise overgrows sensory and sympathetic ganglia.

Necessity, by antibody. Cohen raised an antiserum, serum containing antibodies against one molecule, that blocked NGF’s activity in culture. In companion papers that year, Levi-Montalcini and Booker injected it into newborn mice, with rats and other mammals for comparison. The sympathetic ganglia were destroyed. The procedure became known as immunosympathectomy. The same antiserum has much less effect in older animals, so dependence on NGF is tied to a developmental window. The window differs between populations. In newborn rodents the effect is strong for sympathetic neurons, but NGF-dependent sensory neurons are hit only by earlier exposure: in 1980 Johnson and colleagues destroyed rat dorsal root ganglion neurons with maternal antibody to NGF that reached the fetuses in utero.

An antiserum carries a doubt. It might kill neurons by binding something on them other than NGF.

Necessity, by genetics. A knockout mouse has a gene disrupted so that it makes no working protein, and it settles the antibody doubt. In 1994 Crowley and colleagues reported mice lacking NGF, and Smeyne, Klein, Barbacid and colleagues reported mice lacking a functional TrkA, the NGF receptor. TrkA is a receptor tyrosine kinase of the kind described in that lesson; the receptor family gets its own lesson, that lesson. Both mutants lose sympathetic neurons and many sensory neurons. The sensory losses fall on the neurons that express TrkA, mostly small neurons that sense pain and temperature. Large neurons that sense muscle stretch depend on a different neurotrophin. Basal forebrain cholinergic neurons, which respond to NGF in culture, still develop in NGF-null mice. Not every neuron that can respond to a factor needs it to exist.

The knockout also gives a time course. In 1996 Anne Fagan, Story Landis, Barbacid and colleagues counted the superior cervical ganglion, the sympathetic ganglion at the top of the neck, in TrkA-null mice at a series of ages. Loss of TrkA did not affect neurogenesis, neuronal markers or initial axon growth, but the receptor was absolutely necessary for the survival of sympathetic neurons after embryonic day 15.5. The ganglion formed and then lost its neurons. The neurons were produced; what failed was survival.

The hypothesis stated: neurons compete for a limiting amount

Chris’s statement of it: “Neurons compete for a limiting amount of neurotrophic factor for survival.”

The targets supply it. On his list of NGF-producing targets are the eye, salivary gland, pineal gland, blood vessels, spleen, heart and fat. Sympathetic and sensory axons that reach these tissues take up NGF, and neurons that get too little die by apoptosis, a death that needs the pro-apoptotic protein Bax. Mice lacking Bax keep sympathetic neurons that would otherwise die, a story told in that lesson.

Stated this way, the hypothesis makes predictions you can check one at a time:

• Remove the target, and more neurons die. Hamburger’s limb-bud removals fit.
• Add target, and fewer die. In 1976 Margaret Hollyday and Hamburger grafted an extra leg onto chick embryos. Lumbar motor neuron numbers were not raised before the normal loss, but more remained later in embryos whose grafts moved vigorously.
• Add factor, and fewer die. Remove factor, and more die.
• Death should happen when axons reach the target and begin to compete, not before.

That extra-leg result came from motor neurons, which do not depend on NGF. It supports the general idea of a limiting target resource; it says nothing about which molecule. How a target picks which neurons win is taken up in that lesson.

If you add excess NGF to a developing chick embryo, what happens?

Chris puts this to his class as a poll. The options are no change, more sympathetic neurons surviving than normal, ectopic sympathetic ganglia, or sympathetic neurons invading every tissue. The answer is more survival. The ganglia get larger because more of the neurons that were made escape the normal death. No new neurons appear from nothing, and no neuron changes what kind of neuron it is.

How would you know? A larger ganglion at one late age cannot distinguish survival from extra division. You count neurons and dying cells across the death period and mark dividing cells. In 1981 Hamburger, Brunso-Bechtold and Yip did this kind of counting in chick dorsal root ganglia and found that NGF reduced the normal death.

This gives you a ceiling:

neurons surviving = the smaller of (neurons made) and (neurons the target’s factor can support)

Extra factor raises the second term. Nothing about a survival factor raises the first. Once every neuron has more factor than it can use, a larger dose adds nothing.

The ceiling holds for the population you measured, with its own factor. Motor neurons are the caution. Limb motor neurons do not depend on NGF, and in chick embryos several other factors each reduce part of their normal death. A ceiling found for sympathetic neurons and NGF does not transfer to them.

Permissive or instructive?

NGF passes the tests of necessity and sufficiency for survival. Neither test says whether NGF chooses anything. A permissive signal allows an outcome that something else chooses. An instructive signal chooses the outcome itself. A useful analogy is a stage light: it lets the audience see the play, but the script decides what happens.

Removing a signal and restoring it cannot separate the two, because in both cases the signal is simply present or absent. What separates them is changing how the signal is delivered while keeping it present. Supply it evenly everywhere, or at an arbitrary new place or time, and ask whether the pattern still forms. If it does, something else is reading position, and the signal was permissive. If the pattern follows the delivery, the signal is instructive. In 1985 Kathryn Anderson and colleagues used this logic in fly eggs: wild-type cytoplasm injected into embryos from Toll mutant mothers, which are dorsalized, set the ventralmost part of the rescued pattern at the injection site, wherever the needle went in.

For survival, NGF behaves as permissive: it keeps alive neurons that were already specified as sympathetic or sensory. For other effects it is not merely permissive. NGF promotes axon growth and branching, raises expression of TrkA itself in mouse and rat sympathetic neurons, and in mice lacking both NGF and Bax, sympathetic neurons survive but fail to innervate many of their targets. To our knowledge, whether the pattern of innervation follows the pattern of NGF has not been tested by uniform delivery, so that question is still open.

Design problem

You add excess NGF to a developing chick embryo. Predict the result, and design the control that shows NGF changed survival, not proliferation or fate.