Neurotrophins, Trk Receptors and p75

Add mature nerve growth factor (NGF) to cultured sympathetic neurons and they survive. Add the same protein before an enzyme has trimmed it, and in Barbara Hempstead’s 2001 experiments the neurons died.

The last lesson left you with a rule: a target makes a limited amount of a survival factor, and neurons that fail to get enough die. That rule raises two questions. How does a neuron know which factor is its own? And how can one factor carry opposite messages? The answer to both lies in the receptors: three kinases that each prefer certain factors, and one receptor that binds them all.

The idea: three dedicated kinases and one shared receptor

NGF has three mammalian relatives, so there are four mammalian neurotrophins: NGF, brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3) and neurotrophin-4 (NT-4, called NT-4/5 in older papers). Some fish have extra members (NT-6 and NT-7), which is why the count is given for mammals. Each is a dimer of two identical chains, and all four share the same fold.

Each factor prefers one of three receptors called Trk (pronounced “track”) receptors. A Trk receptor is a receptor tyrosine kinase: the factor dimer brings two receptors together, and their kinase domains phosphorylate each other, as in that lesson. The phosphorylated tyrosines then start two survival branches you have met, the Ras to ERK cascade (that lesson) and the PI3K to Akt route (that lesson).

Neurotrophin	Preferred Trk	Also binds	A mouse population that needs it
NGF	TrkA	p75	sympathetic neurons; small pain- and heat-sensing neurons
BDNF	TrkB	p75	vestibular and nodose-petrosal sensory neurons
NT-4	TrkB	p75	nodose-petrosal sensory neurons
NT-3	TrkC, and more weakly TrkA and TrkB	p75	proprioceptive sensory neurons

The fourth receptor, p75 (also written p75NTR), binds all four factors. It has no kinase. It belongs to the family of the tumour necrosis factor (TNF) receptor, the family of death receptors such as Fas from that lesson, and like them it carries a death domain inside the cell.

Think of the Trk receptors as three locks, each cut for particular keys, and p75 as a doorbell that any key can ring. What the bell does depends on who is home.

The two kinds of receptor also differ in how tightly they hold NGF. Binding studies on chick embryo sensory neurons found two classes of site, with dissociation constants (Kd) of about 20 picomolar and 2 nanomolar. For round numbers, take 10 pM and 1 nM. For simple binding:

fraction of sites occupied = [NGF] / ([NGF] + Kd)

At 10 pM NGF, half of the high-affinity sites are filled and only about 1% of the low-affinity ones. Targets supply small amounts of factor, so the high-affinity sites are the ones that count for survival. Which protein makes them was the puzzle of the 1980s.

”A series of disparate observations”: finding the receptors

Chris Deppmann puts it this way: “The high-affinity NGF receptor was discovered through a series of disparate observations.” The pieces came from neurobiology, from cancer biology and from the search for new factors.

The low-affinity receptor came first. In 1986 Moses Chao and colleagues isolated the gene for the human NGF receptor by transferring human DNA into mouse cells, finding transformants that made the human receptor, and cloning the gene. Eric Shooter’s laboratory reported the rat receptor in 1987. This protein, p75, bound NGF with low affinity. On its own it could not account for the high-affinity sites on neurons.

The kinase came from a tumour. In 1986 Mariano Barbacid’s laboratory described an oncogene from a human colon carcinoma in which part of a tropomyosin gene had fused to the gene for a tyrosine kinase (that lesson). The normal gene, trk, encoded a membrane receptor with no known ligand. In spring 1991 two groups showed that NGF binds the Trk protein and causes its tyrosine phosphorylation: David Kaplan, Barbara Hempstead, Dionisio Martin-Zanca, Moses Chao and Luis Parada, and, separately, Ruediger Klein, Barbacid and colleagues. The protein was renamed TrkA.

That did not end the argument about high-affinity binding. Hempstead and colleagues reported in 1991 that high-affinity NGF binding required TrkA and p75 expressed together. Barbacid’s group found that fibroblasts carrying TrkA still divided in response to NGF without needing p75. The two results measured different things, binding affinity in one and a cell response in the other, and the debate ran for years. A 2022 review concludes that p75 and TrkA form a complex that gives high-affinity NGF binding, but how the two receptors combine is still not settled.

The other factors came from brain. In 1982 Yves-Alain Barde, Deborah Edgar and Hans Thoenen purified BDNF from pig brain as a protein that kept cultured chick sensory neurons alive. Its cloning in 1989 showed that it resembles NGF, and searches for genes resembling both found NT-3 in 1990 and NT-4 in 1991. Klein, in Barbacid’s laboratory, had already cloned a relative of trk, trkB, in 1989. In 1991 three groups showed that TrkB is a receptor for BDNF and NT-3, and Fabienne Lamballe, Klein and Barbacid reported a third relative, TrkC, that prefers NT-3. NT-4 was shown to act through TrkB soon after.

Three deletions, three populations

A knockout mouse is one in which a single gene has been disrupted so that it makes no working protein. Deleting each neurotrophin or Trk gene in mice, one at a time, sorts the family by which neurons are lost.

• NGF and TrkA. Mice lacking NGF or TrkA lose sympathetic neurons and many small sensory neurons that sense pain and heat. In NGF-null mice the basal forebrain cholinergic neurons still develop. In TrkA-null mice, Anne Fagan and colleagues followed the superior cervical ganglion, a sympathetic ganglion in the neck, through development. It formed with a normal number of cells at embryonic day 13.5, then lost about half of them by birth and nearly all of its principal neurons by postnatal day 9. The neurons were made and then died.
• BDNF, NT-4 and TrkB. TrkB-null mice have neuron deficits in both the central and peripheral nervous systems, including motor neurons, and most die within a day of birth. BDNF-null mice lose vestibular neurons, which sense balance and head movement, and many nodose-petrosal neurons, which carry signals from the heart, lungs and gut. NT-4-null mice lose nodose-petrosal neurons too.
• NT-3 and TrkC. Mice lacking NT-3 or TrkC lose proprioceptive afferents, the sensory neurons that report where a limb is, and they move and hold their limbs abnormally.

No two deletions give the same loss. The four factors share a fold and a co-receptor, but to a neuron they are separate signals. Which population depends on which factor tracks which Trk receptor it makes and which factor its target makes. The map has overlaps: NT-3 can activate TrkA and TrkB, and some populations need more than one factor at different stages. In most of these mice the loss in a ganglion is partial, so “lose” means fewer, not none, unless a study counted none.

Dose matters as well. Mice with one copy of the BDNF gene eat more and become obese, and mice with one copy of the NGF gene show shrinkage of basal forebrain cholinergic neurons and poorer memory. Half the factor is not always enough.

Pro-death signaling

The table has a gap. If p75 binds every neurotrophin, what does it do on its own? In the mid-1990s two groups found that it can kill. Barde’s group reported that NGF kills cells in the developing chick retina through p75, and Chao’s group reported that NGF kills cultured rat oligodendrocytes, the myelin-making cells of the brain and spinal cord, which lack TrkA.

Mature NGF, though, binds p75 weakly. The stronger ligand turned out to be the precursor. Like most secreted proteins, NGF is made as a longer chain, proNGF, with an extra stretch at its front called the pro-domain. The accepted view had been that the pro-domain is trimmed inside the cell. In 2001 Ramee Lee, Pouneh Kermani, Kenneth Teng and Barbara Hempstead reported that cells secrete proNGF uncut, and that the cut is made outside the cell by plasmin, a protease from blood, and by particular matrix metalloproteinases. In their experiments uncut proNGF bound p75 tightly, activated TrkA very little, and triggered apoptosis of cultured sympathetic neurons through p75. Mature NGF kept the same kind of neuron alive through TrkA. In 2004 Anders Nykjær, Claus Munck Petersen and colleagues showed that p75 needs a partner, sortilin, which binds the pro-domain, to deliver this death signal.

Keep the scope the papers give. Precursor killing was shown in cultured neurons, in cells of the oligodendrocyte lineage and in rodent injury models. How much it contributes to the normal deaths of development is less clear. What happens after the death signal starts is the subject of that lesson, and how competing neurons use p75 against each other is in that lesson.

What sets a response: the cut, the receptor set and its balance

Put the pieces together and three variables decide what one neurotrophin does at one cell.

1. Whether the ligand has been cut. Uncut proNGF binds p75 with sortilin and can kill. Mature NGF binds TrkA and supports survival. Enzymes outside the cell decide which form arrives.
2. Which receptors the cell makes. A neuron answers the factor whose Trk receptor it makes. Every neuron that makes p75 can also hear every factor through it.
3. How much of each it makes. With sortilin, p75 carries a death signal; with TrkA, it can raise NGF binding. So the ratio of p75 to TrkA changes what the same ligand does.

To change a neuron’s response without changing the gene for the factor, change one of these: give the neuron a different Trk receptor, remove its p75, or change the proteases around it.

Design problem

Your purified “NGF” kills cultured cells that express p75 but little TrkA, while a second batch keeps sympathetic neurons alive. Design the experiments that find out why.

One way to do it

Start with the protein itself. Run both batches on a gel and probe a western blot with an antibody to NGF, and with one to the pro-domain. If the killing batch carries a larger band that the pro-domain antibody recognises, it contains uncut proNGF.

Next, separate the two forms. Make a proNGF with its cleavage site mutated, so no protease can trim it, and test it beside pure mature NGF at the same concentration on both kinds of cell. Prediction: the uncleavable precursor kills the p75 cells and does little for sympathetic neurons, while mature NGF does the reverse.

Then test the receptors. Block p75 with an antibody, or use cells from p75-null mice, and block sortilin, for example with its pro-domain-binding inhibitor or by knocking it down. If killing by the precursor disappears when either partner is gone, the precursor acts through p75 with sortilin.

Last, ask how the first batch came to contain precursor. Add plasmin to it; if the killing activity falls as the precursor band is converted to mature NGF, the difference between the batches was processing, not two different genes.