Interferons and natural killers covers the main trick: a natural killer (NK) cell attacks cells that have stopped showing their MHC flags. This page looks at the cell itself: how it decides, what it does besides killing, what moves its numbers, and what a test of it can tell you.
A lymphocyte with no custom receptor
NK cells are lymphocytes, like T and B cells (see The family of immune cells), but they never build a one-off receptor by shuffling genes. Each carries a fixed set of ready-made receptors, so it can act in the first days of an infection instead of waiting for the right cell to be found and multiplied.
They make up about 5 to 15% of the lymphocytes in blood, and they turn over fast. When dividing cells were labelled in healthy young adults, the labelled NK cells left the blood with a half-life of under 10 days.
Two kinds in the blood
They're told apart by how much of a surface protein, CD56, they carry.
| CD56 dim | CD56 bright | |
|---|---|---|
| Share of NK cells in blood | About 9 in 10 | About 1 in 10 |
| Main job | Killing | Releasing cytokines, above all interferon gamma |
| CD16, the grip for antibodies | Plenty | Little or none |
| Where | Mostly blood | More in lymph nodes and tissues |
How it decides
An NK cell that touches another cell adds up two sets of signals.
- Stop. Receptors that recognise MHC class I, the KIR family and NKG2A. A healthy cell showing its normal flags holds the NK cell back.
- Go. Receptors such as NKG2D, which recognise "stress flags" that a cell puts up when it's infected, damaged or turning cancerous. CD16 is another: it grips the stem of an antibody stuck to a target.
If go outweighs stop, the cell is killed. A virus that strips MHC off the surface removes the stop signal, and enough stress flags can tip the balance even with MHC in place.
There's a safeguard as well. An NK cell is only fully armed if its stop receptors fit your own MHC, a training step called licensing. One that couldn't be told to stop stays weak.
How it kills
The NK cell presses against its target and releases granules into the gap. Perforin opens pores in the target, and granzymes go through them and set off the cell's self-destruct programme. A second, slower route works through "death receptors" on the target.
Releasing the granules needs calcium to flow into the NK cell. In the rare people whose calcium channel ORAI1 doesn't work, NK cells still find their targets and line up their granules, but can't release them.
More than killing
- Interferon gamma. NK cells are an early source of it. It tells macrophages to destroy what they've eaten, and pushes helper T cells towards the programme for germs that live inside cells.
- Something like memory. People who carry cytomegalovirus have more NK cells bearing one receptor, NKG2C, an imprint of that infection. Carrying EBV or herpes simplex doesn't leave the same mark. So an innate cell can be shaped by what it has met, which blurs the line with memory.
- Pregnancy. A distinct kind of NK cell is abundant in the lining of the womb in early pregnancy, where it seems to regulate how the placenta beds in. Exactly how isn't clear.
What they matter for
- Herpesviruses. A teenager with no NK cells, reported in 1989, had severe herpesvirus infections. More than 40 inherited conditions are now known to impair NK cells. See Viruses that stay and Too little.
- Cancer. In 3,625 Japanese adults followed for 11 years, those whose blood cells were in the lowest third for natural killing went on to get more cancer. The middle and top thirds had about 40% less. That's an association from one test years earlier, so it doesn't show that raising NK activity would prevent anything. See Cancer.
What moves the numbers
NK cells are the immune cell most responsive to exercise.
- Exercise and sudden stress pull them into the blood within minutes, through adrenaline. An hour or two after hard exercise, lymphocyte counts dip below where they started. That dip was once read as an "open window" for infection, but the cells seem to be moving into tissues, not disappearing. See Sleep, stress and exercise.
- Sleep. In 42 healthy men, one night kept awake until 3 am lowered NK killing the next day. It was back to normal after a night of recovery sleep.
So a single NK reading says as much about the last few hours as about the person.
Measuring them
| Test | What it does | Where it's used |
|---|---|---|
| NK count | Counts the cells carrying CD56 or CD16 and no CD3, as part of lymphocyte subsets | Routine |
| NK function | Mixes fresh blood cells with a standard target cell and measures how many targets are killed, or how many NK cells release their granules | Specialist labs, for suspected inherited NK deficiency. For HLH, labs now mostly use the granule-release version and a perforin stain |
A count and a function test measure different things: a normal number of NK cells can kill poorly. The function test needs fresh blood, and results differ with the method and the lab. It has a clear use in the rare conditions above. Outside them, no agreed treatment follows from the result, which is why commercial "NK panels" are of little use. See Reading immune blood tests.
NK cells in ME/CFS
Weaker NK killing is the immune finding most often reported in ME/CFS. A 2024 analysis that pooled 28 papers put it at about half the level of healthy controls.
It's less solid than that sounds. The pooled analysis was partly written by the group behind the calcium-channel idea below. It found signs that studies showing no difference had gone unpublished, and its results varied a lot between studies and methods. It also left out one of the largest studies of killing, from the US CDC, because its blood samples had been shipped overnight. That study measured 174 patients across several clinics, and their NK cells killed as well as healthy people's. The largest study of any kind, of 251 patients in the UK ME/CFS Biobank, found no difference in NK numbers, subtypes or their response when stimulated, though it didn't measure killing itself.
A dip in NK killing isn't specific either, since one short night produces one. And the severe herpesvirus infections that come with a true NK deficiency aren't a feature of ME/CFS. Nobody has shown whether weaker killing is a cause, a consequence or a bystander, and it isn't a diagnostic test. See After the infection.
One proposed explanation involves a calcium channel in NK cells. It's the newest of the ideas behind low-dose naltrexone.
An NK cell weighs stop signals against go signals and kills what fails the check, with no need to have met it before. NK cells matter most against herpesviruses. Their numbers in blood swing with exercise, stress and sleep, and a count doesn't say how well they work.