A cytokine is a protein that carries a message. In the US National Cancer Institute's cancer dictionary, cytokines are defined as a type of protein made by certain immune and non-immune cells that has an effect on the immune system: some stimulate the immune system, others slow it down.
The distinction worth holding on to is between a cell and its signal. An immune cell is a sender; a cytokine is the note it releases; the cells that change what they do after receiving it are the readers. Nothing in that chain is interchangeable. A cell is not a cytokine, and a cytokine is not a general verdict that the immune system is up or down.
One laboratory assay shows how the parts line up. A methods paper on functional analysis of human NK cells by flow cytometry describes stimulating NK cells — a kind of immune cell — with target cells, then reading two things by flow cytometry, a technique that measures fluorescent labels on individual cells. The first readout is CD107a, a marker used to detect degranulation, the release of granules from inside the cell. The second is the cytokines the NK cells produce. Trigger, sender, release and measurement all sit inside one experiment.
That paper also makes a point that matters well beyond the experiment: its authors state plainly that degranulation is not the same as lysis of the target cell. A cell can light up with the marker while the target cell survives.
A family of signals, pointing in two directions
Read the NCI definition closely and one detail stands out: cytokines are not all on the same side. Some stimulate the immune system, some slow it down. The word therefore labels a family of protein signals rather than a fixed effect. It is closer to the word "letter" than to the phrase "good news" — it tells you the format of the thing, not what it says.
The definition also names immune and non-immune cells as producers, so cytokines are not a private language spoken only between immune cells.
The implication for anyone reading immunology news is that "cytokines rose" is an incomplete sentence until you know which molecule, made by which cell, and under what conditions.
Production: a signal usually starts with a trigger
In the NK assay, the trigger is built into the design. The cells are stimulated with target cells — the paper uses K562 cells — and only then are the cytokine readouts taken. That choice is a reminder that production is studied as a response to something rather than as an idle hum. Reading the assay that way is an interpretation of the method, not a reported finding; the paper's purpose is to describe a technique, not to announce a result.
The two readouts also answer different questions. Detecting CD107a asks whether the cell released its granules. Measuring cytokines asks whether the cell is making signalling proteins. One assay can do both, but the outputs are not interchangeable, and neither is a clinical outcome.
Transmission: a signal needs something that can read it
Between production and response sits the step that is easiest to skip. The signal has to reach something capable of responding. The NCI definition says cytokines have an effect on the immune system, which implies recipients: cells that can detect the molecule and change their behaviour.
How far a given cytokine travels, by what route, and how long it persists are details the general definition does not settle. They depend on the specific molecule, and one NK-cell assay cannot fill that gap — it records what the NK cell does, not what happens elsewhere afterwards.
Response: a marker is a stand-in, not the event
This is where the NK paper earns its place. Degranulation is often treated as shorthand for killing, and the authors explicitly warn that it is not. A rise in CD107a tells you a cell has been triggered into a release; it does not tell you the target cell died.
That kind of stand-in is called a surrogate marker: a measurable signal used in place of the thing you actually care about. Surrogates are useful because they are easier and often faster to measure. They are also where casual reading goes wrong, because a proxy that moves is not automatic evidence that the underlying event followed.
The same caution applies to cytokine readouts. A fluorescent signal from an antibody stain shows that a cell in that assay, under those conditions, made the protein. It does not by itself show what the same cell would do inside a body, how much would be produced, or in response to what.
Swipe or scroll to read the full table →
| Readout | What it directly shows | What it leaves open on its own |
|---|---|---|
| A cytokine is detected after stimulation | That the cells in that assay produced it under those conditions | Whether the same cells do so in a person, in what quantity, or in response to a real infection |
| CD107a increases | That a marker of degranulation has risen | Whether the target cell was destroyed — the paper's authors draw this line themselves |
| A molecule is called a cytokine | That it belongs to the protein family the NCI defines | Whether it stimulates or restrains an immune response, since the definition covers both |
Four questions to ask about any cytokine claim
- Which molecule? "Cytokine" is a family, and the NCI definition spans both stimulatory and inhibitory members.
- Made by what, and in response to what? Production in the assay follows a supplied trigger; a result with no stated trigger leaves out half the story.
- Measured how? Production, release and effect are three separate measurements, and CD107a shows how easily one gets read as another.
- Measured where? An assay plate is not a person. That last step is my own extrapolation — the paper's caveat concerns lysis, not clinical meaning.
The chain of production, transmission and response is a more useful mental model than a definition on its own. It also explains why the word cytokines rarely appears in a helpful sentence without companions: which one, from which cell, in response to what, and measured how.
This article is general information about immunology, not medical advice for any individual.