Written and maintained by CASRAI Editorial Board
Last updated
A dot plot showing 15% of events in the upper-right quadrant is not, by itself, evidence of 15% apoptosis. Annexin V and propidium iodide (PI) each report on a different membrane event — phosphatidylserine exposure and membrane permeability — and both events can be produced by causes that have nothing to do with programmed cell death: a calcium-chelating wash buffer, a rough dissociation step, or simply reading the plate an hour later than the last one. This guide covers what the two dyes actually detect, the buffer requirement that silently breaks the assay when it’s missed, the controls that let you tell a real apoptotic signal from an artifact, how to read the four resulting populations, and the timing and handling variables that determine whether you’re measuring apoptosis, secondary necrosis, or damage introduced by your own protocol.
What Annexin V and PI Actually Detect
Phosphatidylserine (PS) is a phospholipid that, in a healthy cell, sits almost exclusively on the inner leaflet of the plasma membrane, facing the cytoplasm. One of the earliest committed steps of apoptosis is the loss of that asymmetry: PS translocates to the outer leaflet, where it becomes accessible from outside the cell. Annexin V is a small protein with a high affinity for PS, but only when calcium ions are bound to it — it is not a general apoptosis stain, it is specifically a calcium-dependent PS-binding reagent conjugated to a fluorophore. A cell with externalized PS and an intact, impermeable plasma membrane is the operational definition of early apoptosis in this assay: the cell has committed to the death program, but its membrane has not yet failed.
PI is a nucleic-acid intercalating dye that cannot cross an intact plasma membrane. It is excluded from live cells and from early apoptotic cells for the same reason — the membrane barrier is still functionally intact even after PS has flipped. PI only enters and stains the nucleus once membrane integrity is lost, which happens in late apoptosis (as the cell proceeds toward secondary necrosis) and in primary necrosis, where membrane rupture is the initiating event rather than a downstream consequence. Neither dye on its own can distinguish apoptosis from necrosis; the assay’s entire diagnostic value comes from running both together and reading the combination — a different exclusion principle from the one behind trypan blue viability counting, which reports only membrane rupture and misses early apoptotic cells entirely, since their membrane is still intact.
The Buffer Requirement That Breaks the Assay Silently
Because Annexin V’s affinity for PS is calcium-dependent, the entire staining step has to happen in a binding buffer that contains calcium (typically as CaCl₂), not in standard PBS. This is the single most common way this assay fails without producing an obvious error: if cells are washed or resuspended in a calcium-free buffer, or in one containing a calcium chelator — EDTA or EGTA, both common additions to dissociation buffers and some wash steps — Annexin V will not bind PS even in a genuinely apoptotic sample. The result is a false negative that looks like a clean, healthy population, with no error message and no obvious artifact on the plot. If a sample that should show apoptosis (a positive-induction control, see below) comes back essentially all double-negative, check the buffer chain first, specifically for any EDTA-containing step between harvest and staining, before concluding the induction treatment failed.
Controls: What Has to Run Alongside Every Sample
A single stained sample, read against no reference, cannot be interpreted. At minimum:
- Unstained cells — establishes autofluorescence and where the true double-negative (live) population sits on both axes, before any dye is added.
- Single-stain Annexin V only and single-stain PI only — needed to set compensation between the two channels if the fluorophore pairing has meaningful spectral overlap, and to confirm each reagent is behaving correctly in isolation. Treat this the same as any other multicolor panel: single-stain controls are a compensation requirement, not an optional check, and an FMO does not substitute for one.
- An untreated negative-control population — healthy cells carried through the exact same handling and staining steps as the treated sample, so any signal produced by the protocol itself (see handling-induced damage, below) shows up in the negative control rather than being misread as a treatment effect.
- A positive-induction control — a sample deliberately driven into apoptosis by a treatment independent of the one under investigation (a brief heat shock, UV exposure, or a known pro-apoptotic compound, chosen so it doesn’t share a mechanism with the experimental treatment). Its job is to confirm the assay itself is working — reagents active, buffer correct, gates set sensibly — independent of whether the experimental treatment does anything. Without it, a flat result in the experimental condition is ambiguous between “no apoptosis occurred” and “the assay didn’t work this run.”
Reading the Four Quadrants
On a biaxial plot of Annexin V (x-axis) against PI (y-axis), the four resulting populations are conventionally read as follows:
| Quadrant | Annexin V | PI | Interpretation |
|---|---|---|---|
| Lower-left | Negative | Negative | Live, membrane-intact cells with PS still confined to the inner leaflet. |
| Lower-right | Positive | Negative | Early apoptotic: PS externalized, membrane still impermeable to PI. |
| Upper-right | Positive | Positive | Late apoptotic / secondary necrotic: PS externalized and membrane integrity has now been lost. |
| Upper-left | Negative | Positive | Ambiguous — see below. |
The upper-left quadrant (PI-positive, Annexin V-negative) is the one most often misread. Biologically it is generally interpreted as primary necrosis or mechanical/handling damage — membrane rupture without the preceding PS-externalization step — rather than a stage of apoptosis proper, since the canonical apoptotic sequence externalizes PS before it loses membrane integrity. A sample with a substantial upper-left population alongside a low unstained-control event count is a signal to check your dissociation and handling protocol before attributing anything to the experimental treatment, not to report it as a form of apoptosis.
The Timing Window: Apoptosis vs. Secondary Necrosis
Apoptosis is not a fixed state; a cell that is early apoptotic (Annexin V+/PI−) at the moment of staining will, if left long enough, lose membrane integrity and move into the late apoptotic/secondary-necrotic quadrant even with no further intervention. This means the population distribution you record depends on how long elapses between harvest and acquisition, not solely on the biology of the treatment. Stain and acquire samples within a consistent, short window — stain, then run on the cytometer promptly rather than batching several stained samples on ice for later acquisition — and use the same window across every condition in an experiment. A treatment that looks like it produces more “late apoptosis” than a comparator may simply have been acquired later, not more severely.
Handling-Induced Damage vs. Real Apoptosis
Because the assay reads membrane state directly, any handling step that stresses the plasma membrane can produce a signal that mimics or masks the biology under investigation:
- Enzymatic dissociation. Aggressive or prolonged trypsinization of adherent cells can itself induce PS externalization and membrane stress as an artifact of detachment, independent of the experimental treatment. Where the assay’s outcome is sensitive to this, a gentler or non-enzymatic dissociation method (EDTA-based dissociation, or a shorter enzymatic exposure) reduces the risk of a handling-driven signal — but note that EDTA-containing dissociation buffers must be fully washed out and replaced with a calcium-containing binding buffer before staining, per the buffer requirement above.
- Mechanical stress. Vigorous pipetting, vortexing, or excessive centrifugal force during wash steps can rupture a fraction of cells outright, producing PI-positive/Annexin V-negative events that have nothing to do with the treatment.
- Why the untreated negative control matters here. Any of these handling artifacts will also appear in the untreated control, since it goes through the identical dissociation and wash steps. A negative control with an unexpectedly high apoptotic or necrotic background is the signal that the protocol, not the biology, needs attention — check it before interpreting the treated condition.
Gating Before You Ever Reach the Annexin V/PI Plot
The Annexin V/PI biaxial plot should be gated on single cells after debris exclusion, following the same forward-scatter/side-scatter debris gate and singlet-discrimination gate used in any standard flow cytometry workflow — debris and doublets both distort apoptosis quadrant percentages, and a debris-inclusive gate will inflate the apparent double-negative or PI-positive population depending on where the debris happens to fall. If compensation is needed between the Annexin V fluorophore and PI’s channel, set it up using single-stain controls before setting the apoptosis gates, not after.
Frequently Asked Questions
Can Annexin V/PI staining be used on fixed cells?
PI’s exclusion from intact membranes is the basis of the necrosis/late-apoptosis readout, and fixation permeabilizes the membrane for essentially all cells regardless of their apoptotic state — so a standard Annexin V/PI protocol is designed for live, unfixed cells and is not meaningful after fixation. If a fixable readout is needed, that calls for a different reagent set (e.g., a fixable viability dye or a different apoptosis marker), not this assay run on fixed material.
Why did my positive-induction control fail to show any apoptotic signal?
Check the buffer chain first, specifically for any calcium-chelating step (EDTA/EGTA) between harvest and the Annexin V staining step — a calcium-free environment silently prevents Annexin V from binding PS even when the induction treatment worked as intended, and produces no error, just a falsely clean-looking plot.
What’s the difference between using PI and using 7-AAD as the second dye?
Both are membrane-impermeant nucleic-acid dyes excluded by intact membranes and both pair with Annexin V on the same logic, but they occupy different emission windows — PI is broadly emitting and can be difficult to combine with other channels in a larger panel, while 7-AAD’s narrower emission profile makes it easier to slot into a multicolor panel alongside additional markers. The choice is largely a panel-design decision, not a difference in the underlying apoptosis logic.
How long after staining should the sample be acquired?
As soon as practical, and consistently across every condition in the experiment. Because early apoptotic cells continue progressing toward membrane permeabilization over time, a delay between staining and acquisition shifts events from the early-apoptotic to the late-apoptotic/necrotic quadrant independent of the treatment, so an inconsistent delay between conditions will read as a treatment effect that isn’t real.
Is a high percentage in the PI-only (upper-left) quadrant something to report as apoptosis?
No — that quadrant is generally interpreted as primary necrosis or handling/mechanical damage rather than a stage of programmed cell death, since it shows membrane rupture without the PS-externalization step that defines the apoptotic sequence. A substantial upper-left population is a prompt to review dissociation and handling technique, and to check the untreated negative control for the same signal, before attributing it to the experimental condition.
