The Death Machinery in Neurons

Take nerve growth factor (NGF) away from a young sympathetic neuron in a dish, and it dies within a day or two. Add a drug that stops it from making new proteins, and the death is delayed.

The second sentence is the strange one. A cell starved of what it needs should simply run down, and stopping protein synthesis should, if anything, make it run down faster. Instead the neuron has to make something in order to die. In 1988 David Martin, Eugene Johnson and colleagues reported that inhibitors of RNA synthesis or of protein synthesis kept rat sympathetic neurons alive after NGF was withdrawn. The previous lessons showed that developing nervous systems lose a large share of their neurons. This one asks what machinery carries out those deaths, where it was found, and what you learn when you take it away.

Mechanisms of Apoptosis: death that takes work

Cells can die in two broad ways. In necrosis, after trauma or loss of blood supply, the cell swells, its membranes fail and its contents spill out, inflaming the tissue around it. In apoptosis, a term introduced by John Kerr, Andrew Wyllie and Alastair Currie in 1972, the cell shrinks, its chromatin condenses into dense clumps, and it breaks into membrane-wrapped pieces that other cells swallow. A nucleus condensed this way is called pyknotic, and pyknotic nuclei were what researchers counted in early studies of neuron death in embryos.

Apoptosis is carried out by proteases called caspases. In this course’s lecture the name is unpacked as “Cysteine requiring Aspartate Protease”; the standard expansion is cysteine-dependent aspartate-directed protease. Both say two things: the enzyme cuts with a cysteine in its active site, and it cuts its targets just after an aspartate. Caspases are made as inactive precursors. Initiator caspases switch on when they are gathered onto a platform, and they then cut and activate executioner caspases, which cut hundreds of cellular proteins. The Cell Signaling course teaches the biochemistry in that lesson: proteins of the BCL-2 family decide whether mitochondria release cytochrome c, and released cytochrome c sets off the caspase cascade.

Think of it as a demolition crew kept on retainer. The crew is already on site, with its tools locked away; the survival signal pays it to wait. The Martin experiment says that when the payments stop, the cell still has to issue a work order, in the form of new RNA and protein, before the crew starts. The next question was who the crew members are, and the clearest answer came from a worm.

The cells marked to die: a program found in a worm

The nematode Caenorhabditis elegans is about a millimetre long and transparent. With Nomarski optics, which make nuclei visible in a living animal without staining, John Sulston and H. Robert Horvitz followed the cell divisions after hatching, outside the gonad, and in 1977 published the postembryonic cell lineage, the full family tree of those cells. Sulston and colleagues completed the embryonic lineage in 1983. Some branches ended in a death, and the same identified cells died in every animal, at the same place in the tree. Counting both lineages, 131 of the 1,090 somatic cells produced in the hermaphrodite die during development.

A death that happens to a named cell on schedule is a phenotype you can screen for. Hilary Ellis and Horvitz started from a mutant, ced-1, in which corpses are not engulfed and so stay visible, then searched for new mutations in which the corpses never appeared. In 1986 they reported two genes, ced-3 and ced-4, that are needed for programmed cell death. In strong mutants the cells that should have died lived, and some took on features of neurons, such as making serotonin or dopamine. The animals looked grossly normal in the tests used.

Then came the gene that works the other way. In 1992 Michael Hengartner, Ronald Ellis and Horvitz reported that ced-9 protects cells: a gain-of-function mutation kept doomed cells alive, and loss of ced-9 killed cells that should have lived. Removing ced-3 or ced-4 stopped those extra deaths, so in genetic order CED-9 acts upstream, as a brake on the two killers.

Cloning connected the worm to mammals. In 1993 Junying Yuan, Horvitz and colleagues reported that ced-3 encodes a protein similar to mammalian interleukin-1β-converting enzyme, a protease of the family later named caspases. In 1994 Hengartner and Horvitz reported that ced-9 encodes a relative of human Bcl-2, a protein that keeps cells alive. CED-4 later proved related to mammalian Apaf-1, described by Xiaodong Wang’s laboratory in 1997.

So the core has worm counterparts: the caspases, including caspase-9 and caspase-3, for CED-3; Apaf-1 for CED-4; Bcl-2 and its relatives for CED-9. The match is imperfect. There is no clear role for cytochrome c in worm apoptosis, and the worm has no clear counterpart of Bax. In 2002 Sydney Brenner, Horvitz and Sulston shared the Nobel Prize in Physiology or Medicine for discoveries concerning the genetic regulation of organ development and programmed cell death.

Inside a deprived neuron: from a lost signal to the apoptosome

Return to the sympathetic neuron that has lost NGF. The program can be followed step by step, and each step has been tested by blocking it.

1. The survival signal falls. NGF acting through its receptor TrkA (that lesson) keeps kinases such as Akt active, and Akt holds back several pro-death proteins (that lesson). Without NGF that restraint weakens.
2. A stress pathway turns on. The kinase JNK becomes active and phosphorylates the transcription factor c-Jun (that lesson). In the mid-1990s groups including Johnson’s reported that c-Jun rises in deprived sympathetic neurons and that blocking its function protects them. Among the genes then induced are pro-death BCL-2 family members that carry only a BH3 domain, such as Bim. This is part of the new gene expression that the 1988 inhibitors blocked.
3. Bax opens the mitochondria. In 1996 Thomas Deckwerth, Johnson, William Snider, Stanley Korsmeyer and colleagues reported that sympathetic neurons from mice lacking Bax survive NGF deprivation. Bax is needed for cytochrome c to leave the mitochondria of these neurons.
4. The apoptosome forms. Cytochrome c in the cytoplasm binds Apaf-1, which assembles into a wheel-shaped complex called the apoptosome. The apoptosome activates caspase-9, which cuts and activates caspase-3.
5. The cell is marked for removal. Executioner caspases dismantle the cell and expose an “eat me” signal on its surface, the subject of the last section.

Whole-animal genetics agrees in part. Mice lacking caspase-3 or caspase-9 (knockouts, in which a gene is deleted from the germ line) can have enlarged, malformed forebrains with extra cells. In caspase-3 knockouts the severity depends heavily on the genetic background, the mouse strain the deletion is bred into, and on the C57BL/6J strain the brain develops normally. Caspases are also used locally. As the degeneration lecture puts it, “Caspases required for developmental degeneration but not injury,” a point taken up in Pruning Without Dying.

Design problem

You inject cytochrome c into the cytoplasm of cultured sympathetic neurons kept in NGF, and into neurons that have been deprived of NGF for a day. Predict both results, and say what they reveal about how NGF keeps a neuron alive.

”Running on Surplus”: what Bax-null mice show

If Bax is the gate, what happens in an animal that lacks it? In 1998 Fletcher White, Snider, Korsmeyer and colleagues reported that naturally occurring death was largely eliminated in many neuron populations of Bax-null mice. Across these and later studies, Bax deletion rescues many developing populations, among them sympathetic and sensory neurons, facial motor neurons and retinal ganglion cells. It does not rescue every developing neuron.

The saved neurons were not normal. In several nerves of Bax-null mice, including the facial and optic nerves, axon counts rose by 24 to 35 percent, and many of the extra axons were very small in cross section, a sign that the neurons were not normal. Later work from Ronald Oppenheim’s laboratory found that the extra spinal motor neurons of Bax-null mice had begun to innervate their targets, after which a subset shrank and lost target contact. A neuron saved from death without enough target support survives as a smaller cell.

Natalia Glebova and David Ginty used Bax deletion to ask a different question. Mice lacking NGF lose their sympathetic neurons, so you cannot ask what else NGF does for them. In mice lacking both NGF and Bax, reported in 2004, the sympathetic neurons survived, but their axons failed to reach many targets, and the failure differed by organ. Innervation was absent in some organs, such as the salivary glands, greatly reduced in others, such as the heart, somewhat reduced in a few, such as the stomach and kidneys, and normal in some, such as the trachea. NGF is therefore needed for axons to grow into many targets as well as for survival, and the need differs from organ to organ.

The logic is general. A population that survives because the death machinery was deleted shows you what the death was for, by showing what is still wrong once dying is no longer an option. How targets choose among their neurons is the subject of competition.

Cleaning up: the corpse sends a signal

A developing brain kills many cells, yet at any moment few dead cells are visible, because they are removed quickly. In the worm, neighbouring cells engulf the corpses, which is why the ced-1 mutant, with its lingering corpses, made such a useful screen. In the vertebrate nervous system the main phagocytes are microglia, the brain’s resident immune cells.

The dying cell helps. A healthy cell keeps the lipid phosphatidylserine on the inner face of its plasma membrane. In the cells where this has been studied, caspases cut the enzymes that hold it there and activate a scramblase that mixes the two faces, so phosphatidylserine appears outside, where receptors on phagocytes recognize it as an “eat me” signal. The same exposure lets researchers label apoptotic cells with annexin V, a protein that binds phosphatidylserine. Because the cell is eaten before its membrane fails, its contents never spill, and the tissue is not inflamed. Microglia also remove synapses, which another lesson covers.