# Cell Death and Clearance

> Death is a normal, genetically controlled part of building a body: in a worm, the same 131 cells die on schedule, yet in healthy tissue dead cells are cleared so fast that corpses are rarely seen. One idea ties the lesson together: in apoptosis the plasma membrane stays intact early, often until a phagocyte eats the cell, while necrotic deaths generally open it early and spill. The lesson follows the worm genetics that showed the cell's own genes do the killing, the experiments behind each clearance step (find-me, eat-me, engulfment), the necrotic deaths that are programs too, and how annexin V, propidium iodide, TUNEL, caspase and LDH assays each report one event that more than one kind of death can produce. The way a dying cell looks does not by itself tell you which machinery killed it.

Course: Cell Biology (Division and cell loss). Level: core university (third year). About 25 minutes.
Lesson page (link this one when citing): https://stemhop.com/lessons/cell-death/

Builds on: Membrane Organization and Topology · Vesicle Budding, Targeting and Fusion

## What you'll learn

- Explain how the invariant C. elegans lineage and the ced-3, ced-4 and ced-9 mutants showed that developmental death is genetically programmed.
- Distinguish apoptotic from necrotic morphology, and name necroptosis, pyroptosis and ferroptosis by the machinery each depends on.
- Trace clearance from find-me signals such as nucleotides released through PANX1, to phosphatidylserine exposed by Xkr8, to tethering by Tim4, engulfment by MerTK and digestion in lysosomes.
- Explain why annexin V, propidium iodide, TUNEL, cleaved caspase-3 and LDH release are each ambiguous alone, and combine them over a time course.

> **The one idea to carry through this lesson.** In apoptosis the cell’s own caspases take it apart while its plasma membrane is still intact, which gives a phagocyte time to find and eat it before anything leaks. Necrotic deaths generally open the membrane early and spill. Its limits: an uneaten apoptotic cell often leaks later (secondary necrosis), and some tearing deaths are genetic programs too.

The first film showed what happens; the rest of the page asks how we know, one piece of its idea at a time. The worm shows who takes the cell apart, clearance experiments show what happens while the membrane holds, necrotic programs show what opens it early, and two dyes tell these apart in a dish.

## The worm shows the cell kills itself

The evidence that the cell’s *own* caspases do the killing came from a worm. In the hermaphrodite *Caenorhabditis elegans*, development makes 1,090 somatic cells and kills 131 of them, leaving 959 in the adult (males have a different lineage and count). In living transparent embryos every division can be followed, and the lineage and its fates, including which cells die, are highly invariant. But an invariant pattern says death is reproducible, not that a gene causes it: a cell might die because it is squeezed or starved at the same place every time. The test is genetic: find mutants in which those cells fail to die.

Hilary Ellis and H. Robert Horvitz did that. Their 1986 paper reported that wild-type function of two genes, *ced-3* and *ced-4* (*ced* for cell death abnormal), is needed for the deaths they examined.

**Predict.** Suppose all you knew was that worms lacking *ced-3* show fewer cell corpses. Does that show the cells did not die? What observation would?

> **Check your prediction.** No. Fewer corpses alone could mean corpses were cleared too fast to count (the corpse-count problem below puts numbers on this). Ellis and Horvitz saw more: cells normally fated to die survived, and some differentiated into recognizable cell types. A living, differentiated cell shows the death itself did not happen.

*Figure (AI-generated illustration).* In worms lacking ced-3 or ced-4, cells normally fated to die survived, and some differentiated. A missing corpse could mean fast clearance; a living, differentiated cell shows the death did not happen. Image: a row of cells in which one is fated to die. In the wild type it shrinks and disappears. When ced-3 or ced-4 is lost it survives and differentiates, shown in gold. A small inset contrasts an empty gap, which could mean fast clearance, with a living differentiated cell.

A third gene acted the other way. In 1992 Michael Hengartner, Ronald Ellis and Horvitz reported that an activating mutation in *ced-9* prevented the normal developmental deaths, while inactivating mutations made cells that normally live die.

Geneticists order genes like these by **epistasis**: in a linear pathway, when two loss-of-function mutations have opposite effects, the [double mutant](https://stemhop.com/methods/epistasis-analysis/) usually shows the effect of the gene that acts later.

**Predict.** If *ced-3*, the killer, acts after *ced-9*, the protector, what does a worm lacking both do?

> **Check your prediction.** It lives and keeps the cells that normally die, as the real double mutant does. The killing that follows loss of *ced-9* needs *ced-3*, so *ced-3* acts downstream, carrying out the death that *ced-9* normally prevents.

Ordering of this kind placed a fourth gene, *egl-1*, upstream of *ced-9*, then *ced-4*, then *ced-3*. When the genes were cloned, CED-3 proved to be a caspase, CED-4 resembled human Apaf-1 and CED-9 resembled BCL-2 (Going deeper has a caution about how CED-9 works). In human cells this machinery runs as a cascade: in the intrinsic route BAX and BAK permeabilize the mitochondrial outer membrane, cytochrome c escapes, and caspases (proteases that cut after aspartate) are activated in turn. You do not need those names here: [death receptors, BCL-2 and caspases](https://stemhop.com/lessons/death-receptors-bcl2-caspases/) teaches them, [stress signals](https://stemhop.com/lessons/stress-signals/) covers how p53 can push a damaged cell toward this death, and for the nervous system see [naturally occurring death](https://stemhop.com/lessons/naturally-occurring-death/) and [death machinery in neurons](https://stemhop.com/lessons/death-machinery-in-neurons/).

## Clearance: separating each step from the death

Each clearance step was shown by blocking it without stopping the death. Clearance of apoptotic cells is often called efferocytosis.

**Find me.** The first film’s channel is PANX1. In apoptotic Jurkat cells (a human T-cell line), the effector caspases 3 and 7, the ones that dismantle the cell, cut PANX1 and open it. ATP and UTP leave through it and attract monocytes, blood cells that can become macrophages. Knocking PANX1 down cut nucleotide release without slowing caspase activity, so the signal can be separated from the death. Nucleotides are one of several find-me signals; dying cells also release the lipids lysophosphatidylcholine and sphingosine-1-phosphate and the chemokine CX3CL1.

**Eat me.** Valerie Fadok, Peter Henson and colleagues reported in 1992 that macrophages recognize exposed phosphatidylserine (PS) on apoptotic lymphocytes. PS is normally kept on the inner leaflet, the inner layer of the plasma membrane (see [lipids](https://stemhop.com/lessons/lipids/) and [membrane organization and topology](https://stemhop.com/lessons/membrane-organization-topology/)). The first film’s caspases cut the membrane proteins that set which face PS sits on, and there are two kinds. A **flippase** is a pump that keeps returning PS to the inner face. A **scramblase** is a door that lets lipids mix between the two faces. Caspases cut and activate the scramblase Xkr8, and in mouse cells lacking Xkr8, apoptotic cells failed to expose PS and were engulfed poorly.

**Predict.** Caspases also cut the flippase ATP11C, and cutting switches it off. A cell carries an ATP11C that caspases cannot cut, so the pump keeps running, while Xkr8 is switched on as usual. Does PS reach the outer face during apoptosis?

> **Check your prediction.** No. Cells carrying a caspase-resistant ATP11C did not expose PS during apoptosis. With the pump still running, the door alone is not enough. Exposure needs both steps: the scramblase on and the flippase off.

**Hold, then swallow.** The first film used the simplest picture, a single gulp; in resident mouse peritoneal macrophages it takes two receptors. Tim4 binds PS directly and tethers the dying cell. MerTK, a receptor tyrosine kinase (it signals by phosphorylating proteins on tyrosine), binds PS through the bridging proteins Gas6 or Protein S and drives the engulfment. The phagosome then fuses with lysosomes, which digest the cell, as in [endocytosis](https://stemhop.com/lessons/endocytosis/).

*Figure (AI-generated illustration).* Clearance in steps: nucleotides released through PANX1 call a phagocyte, Xkr8 exposes phosphatidylserine, and Tim4 tethers while MerTK drives engulfment (shown in resident mouse peritoneal macrophages). Image: a numbered plate of clearance. An apoptotic cell releases nucleotides through a PANX1 channel and exposes phosphatidylserine through Xkr8; a macrophage tethers it with Tim4 and engulfs it with MerTK.

### Design problem

Mouse resident peritoneal macrophages are mixed with apoptotic cells. Macrophages lacking Tim4 bind fewer apoptotic cells. Macrophages lacking MerTK bind them normally but take few inside. Design an assay that separates binding from engulfment, and say what each receptor does. (Hint: you need one readout for “stuck to the outside” and another for “inside”.)

> **One way to do it.** Score bound and internalized targets separately. Label targets with a pH-sensitive dye that brightens in acidic compartments, so targets inside acidified phagosomes glow and surface-bound ones do not (acidification takes time, so a target engulfed moments ago may still be dim), and count total contacts by microscopy. Or quench or strip the surface signal after incubation. Run a parallel incubation at 4 °C, where binding still occurs but uptake stops.
>
> Controls: mask PS on the targets with annexin V, which should block both binding and uptake; use live, non-apoptotic targets, which should be ignored. Then put each receptor alone, and both together, into a cell line that does not normally eat cells, and ask which confers binding and which confers engulfment.
>
> The expected answer from Nishi and colleagues (2014) is two steps in this macrophage population: Tim4 tethers, MerTK engulfs, and cells expressing both engulf strongly. Other mouse macrophages depend on Tim4 to different degrees (Yanagihashi and colleagues, 2017), so the result should not be generalized beyond the cells tested.

### Worked problem: why you rarely see a corpse

The numbers are illustrative. At steady state, corpses on view = death rate × time each stays visible. If a tissue replaces 1% of its cells a day, 10,000 per million die daily, about 417 an hour; if each corpse is visible for 1 hour, about 417 per million (0.04%) are on view.

**Predict.** Clearance slows, so each corpse stays visible for 12 hours. The death rate does not change. How many corpses do you see now?

> **Check your prediction.** About 417 × 12 = 5,000 per million, or 0.5%: twelve times as many corpses with no extra death. So a corpse count cannot tell faster clearance from death that failed, which is why the *ced-3* evidence had to be living, differentiated cells.

**Why it matters: inflammation.** The first film’s alarm signals are spilled molecules, such as the protein HMGB1, that can call in immune cells, so apoptosis is usually the quieter death. Usually, not always: apoptosis and its clearance are not always immunologically silent; outcomes depend on context, as the lesson on [TLRs and NF-κB](https://stemhop.com/lessons/tlr-and-nf-kb/) helps explain. Phagocytes differ too: microglia also clear dying cells and pruned synapses (see [microglia and complement pruning](https://stemhop.com/lessons/microglia-complement-pruning/)).

That quieter-or-louder contrast was taught for decades as programmed versus accidental. The second film picks up here: what a torn cell’s look can and cannot tell you about what killed it.

**Film transcript: A Look Is Not a Cause**

### Film transcript: A Look Is Not a Cause

This cell tore open. What killed it: an accident, or a program? Chapter one showed two endings: an apoptotic cell held its membrane until eaten; a necrotic cell tore early. For decades, these two looks were taught as a pair. Apoptosis, named in 1972 by Kerr, Wyllie and Currie, was the programmed death. Necrosis was the accident. That pairing has not survived. A regulated death runs on dedicated machinery.

Genes or drugs can turn that machinery up or down. Such a cell can still look necrotic. Three such deaths are defined by what they depend on. First, necroptosis. Activated RIPK3 adds phosphate to MLKL. Clusters of MLKL then move to the plasma membrane.

Second, pyroptosis. Gasdermins are cut, freeing fragments that form pores in the plasma membrane. It often, but not always, follows inflammatory caspases. Third, ferroptosis, driven by severe, iron-dependent damage to membrane lipids.

GPX4 is the main guard, reducing damaging lipid hydroperoxides. In cultured cells, drugs that trap iron block it. Here is the idea to carry. A death's look names an endpoint, not a cause. To find the cause, block the machinery and see whether the death changes. Now take dying cells in a dish, with no phagocyte. Two dyes read chapter one's two membrane events. Annexin V binds phosphatidylserine it can reach from outside. Propidium iodide, a DNA dye, is kept out by an intact membrane. A flow cytometer reads both colours on each cell and places it in one of four boxes. An apoptotic cell in a dish often becomes flagged while intact. After secondary necrosis, it is flagged and leaky.

The boxes are staining states, not verdicts. Other markers each report one event. TUNEL labels free DNA ends, whatever broke them. Cleaved caspase-3 shows the protease switched on, not that it killed. LDH, a cytosol enzyme, leaks through a torn membrane. So does LDH mean a cell died?

Researchers removed NINJ1, a small plasma membrane protein, from mouse macrophages. Then they triggered death. Will LDH pour out? Very little came out. The cells died, but they stayed swollen, like balloons that never popped. Death and the final tear are separate steps. So LDH measures the tear, not the death. So back to the torn cell. Its look tells you how it ended, not what killed it. To find the killer, block each candidate mechanism and watch what changes.

From the film

- Some deaths that tear the membrane are genetic programs, not accidents.
- In the mammalian model, necroptosis needs RIPK3 and MLKL; pyroptosis needs pore-forming gasdermins; ferroptosis is driven by iron-dependent lipid damage.
- A death's look names an endpoint, not a cause; to find the cause, block the machinery and see whether the death changes.
- Annexin V reads phosphatidylserine it can reach from outside, and propidium iodide reads a leaky membrane.
- Mouse macrophages lacking NINJ1 died but released little LDH, so LDH reports the final tear, not the death.

## Necrotic deaths can be programs too

The second film named the three programs; this section adds what it left out. Kerr, Wyllie and Currie described the apoptotic look in many tissues in 1972: chromatin condenses against the nuclear envelope, and blebs can pinch off as membrane-bound pieces. **Necrosis** shows swollen organelles and a ruptured membrane.

*Figure (AI-generated illustration).* Apoptosis before the membrane fails (left) and necrosis (right) as they look under the microscope. These looks name an endpoint, not a cause: a cell can die with necrotic looks through dedicated machinery. Image: two cells side by side. The apoptotic cell is small, with a fragmented nucleus, condensed chromatin, surface blebs and an intact membrane. The necrotic cell is swollen, with swollen organelles and a ruptured membrane spilling its contents.

A death is **regulated** when it relies on dedicated molecular machinery that genetic changes or drugs can alter, and it can still end with necrotic morphology. The three regulated deaths, in fuller form:

- **Necroptosis** needs RIPK3 and MLKL in the mammalian model: *activated* RIPK3 phosphorylates its substrate MLKL, and the MLKL oligomers then move to the plasma membrane.
- **Pyroptosis** needs gasdermins: cutting frees the fragments that form pores, and often, but not always, follows activation of inflammatory caspases.
- **Ferroptosis**: the iron-dependent damage is severe lipid peroxidation (oxidative damage to membrane lipids). GPX4 holds it off by reducing lipid hydroperoxides; parallel systems such as FSP1, which works through coenzyme Q10, also do. In cultured cells it is blocked by iron chelators and lipophilic antioxidants.

The second film’s test, perturb the machinery and see the death change, is the logic of [experiments that distinguish](https://stemhop.com/lessons/experiments-that-distinguish/), applied to a dying cell.

## Reading the flag and the leak with two dyes

The second film introduced the two dyes; here is how to read them in a dish, where no phagocyte eats the cells.

- **Annexin V reads the flag.** It is a calcium-dependent protein that binds phosphatidylserine (PS), as Tim4 does on a macrophage. Binding means PS is reachable from outside.
- **[Propidium iodide](https://stemhop.com/methods/apoptosis-assays/) (PI) reads the leak.** It is a DNA dye that an intact membrane keeps out. PI entry means the membrane is leaky, whatever made it so.

A flow cytometer reads the fluorescence of single cells streaming past a laser and places each in one of four boxes:

|  | PI negative: membrane intact | PI positive: membrane leaky |
| --- | --- | --- |
| **Annexin V negative: no PS seen outside** | neither signal (AnnV−PI−) | leaky, no flag seen (AnnV−PI+) |
| **Annexin V positive: PS accessible to annexin V** | external PS detected, membrane PI-impermeable (AnnV+PI−) | annexin-accessible PS, membrane PI-permeable (AnnV+PI+) |

The boxes are staining states, not verdicts on life or death: a cell with neither signal may already be committed to die, and some living cells expose PS too.

**Predict.** An apoptotic cell in a dish with no phagocyte starts in flagged, intact and, after secondary necrosis, ends in flagged, leaky. Where does a cell sit that tore open in the first minutes of a necrotic death? Can one time point tell the two apart?

> **Check your prediction.** The torn cell lets PI in, and annexin V gets inside and finds PS on the inner face. So it goes straight to flagged, leaky. **Both deaths end in the same box**; one time point cannot tell them apart. The flagged, intact box fits the apoptotic order but does not prove it: necroptotic cells can also expose PS before they leak, and how long cells stay there must be measured in each system.

### Worked problem: does this death need caspases?

Made-up data: 10,000 Jurkat cells per sample, counted as neither / flagged, intact / flagged, leaky / leaky only.

- Untreated: 9,300 / 400 / 200 / 100
- Drug, 6 hours: 5,800 / 2,900 / 1,000 / 300
- Drug plus a pan-caspase inhibitor (a drug that blocks many caspases at once), 6 hours: 8,600 / 600 / 500 / 300

Each row adds up to 10,000. Assume the inhibitor does not affect the cells on its own, and that the untreated sample is the baseline for every condition.

**Step 1. How much of the flag needed caspases?** Annexin-positive cells (both flagged boxes): 6%, 39% and 11%. The excess is 33 points with the drug and 5 with the inhibitor.

> inhibitor-sensitive share of the drug-induced annexin-positive excess = (33 − 5) / 33 = 28 / 33 = 0.85, about 85%

That is a share of the PS-exposing cells, not of the death. In apoptosis, PS exposure itself needs caspases, so the inhibitor can lower the annexin signal even in cells that still die. The remaining 5 points are caspase-independent PS exposure or incomplete inhibition; these data cannot tell which.

**Step 2. Is the pattern flag-first?** With the drug, the excess is 2,500 flagged, intact and 800 flagged, leaky, about three to one; with the inhibitor it is 200 and 300. That shape is compatible with a flag-first death, but one snapshot cannot show cells moving between boxes, since the ratio also depends on how fast cells enter each box and how long they stay; repeated time points or single-cell tracking would test it. The leaky-only box (200 above baseline in both treated samples, unchanged by the inhibitor) means nothing until you check that its events are whole cells, not debris.

**Step 3. What 6 hours cannot tell you.** Under these assumptions, most of the drug-induced annexin signal is inhibitor-sensitive, which is compatible with caspase-dependent PS exposure but does not by itself establish caspase dependence or show that caspases did the killing. The inhibitor may only *delay* death, or *reroute* it: in cells that make RIPK3, blocking caspase-8 can open the necroptosis route. So follow PI uptake or viability to, say, 24 hours, with and without a RIPK3 or MLKL inhibitor, and add a direct caspase readout and a test of whether the cells can still recover and grow.

*Interactive on the lesson page:* Annexin V and PI quadrants: what share of the death needs caspases?

The other common markers each report one event that can occur in more than one kind of death:

- **TUNEL** labels free DNA ends in fixed cells. In rat liver it labelled necrotic and autolytic (self-digested after death) cells as well as apoptotic ones.
- **Cleaved caspase-3** on an [immunoblot](https://stemhop.com/methods/western-blot/) shows the protease was activated, not that the cell died or that caspase-3 killed it.
- **LDH release** into the medium reports the final rupture that lets out large molecules such as the cytosolic enzyme lactate dehydrogenase (LDH). In mouse macrophages lacking NINJ1, a small plasma membrane protein, cells died yet stayed swollen and released little LDH, showing that rupture is its own step.

So the first film’s idea holds, with its limits: an apoptotic membrane stays intact early, often until the cell is eaten. Each marker reports a state that more than one death can reach, in no fixed order, so combine markers over time, then perturb.

## Going deeper

### The hermaphrodite lineage counts

John Sulston, Einhard Schierenberg, John White and Nichol Thomson followed every embryonic division under a light microscope. They reported in 1983 that 113 cells die during embryogenesis in the hermaphrodite; the rest of the 131 die after hatching.

### From worm genes to the human cascade

In a human [cell-free](https://stemhop.com/methods/in-vitro-reconstitution/) system, Li and colleagues showed in 1997 that cytochrome c and dATP let Apaf-1 bind and activate caspase-9, which then cleaves and activates caspase-3.

The textbook cartoon has CED-9 simply holding CED-4 until EGL-1 releases it. That is a simplification: the genetic phenotypes stand, but the mechanism by which CED-9 protects is less settled. A 2024 study of CED-9 mutants that cannot bind CED-4 found that the binding was needed for a pro-apoptotic role of CED-9 but not for its protective role, in their assays.

---

From STEMhop (https://stemhop.com), free lessons by Christopher Deppmann. Lesson text © the authors. Practice questions: https://stemhop.com/lessons/cell-death/
