Natural Killer Cells

Why natural killer cells matter

• First responders that need no priming. Adaptive immunity is powerful but slow — cytotoxic T cells take 5 to 7 days to be primed, expand, and reach a site of infection. NK cells are pre-armed with lytic granules and kill within hours of encountering a stressed or infected cell, buying time before the adaptive response arrives.
• They close the MHC-loss escape hatch. Cytotoxic T cells only see targets that display peptide on MHC class I. Many viruses (herpesviruses, HIV) and roughly a third to a half of tumors downregulate MHC class I precisely to blind T cells. NK cells turn that camouflage into a red flag: no MHC class I means no inhibition, so the NK cell attacks.
• Tumor immunosurveillance. High peripheral NK activity in an 11-year Japanese cohort study (Imai et al., Lancet 2000, ~3,600 residents) was associated with lower subsequent cancer incidence; low activity tracked with higher risk. NK infiltration into solid tumors is a favorable prognostic marker across several cancer types.
• They deliver antibody therapy. Through CD16 (FcγRIIIa), NK cells are the dominant in-vivo effectors of antibody-dependent cellular cytotoxicity. Rituximab (anti-CD20), trastuzumab (anti-HER2), and cetuximab (anti-EGFR) all work in part by flagging tumor cells for NK-mediated destruction.
• Congenital NK deficiency is dangerous. People with rare inherited NK-cell deficiencies (for example mutations in MCM4, GATA2, or FCGR3A) suffer severe, recurrent, often fatal herpesvirus infections — CMV, VZV, EBV, HSV — underscoring that NK cells are indispensable for antiviral defense, not merely redundant.
• The engine of pregnancy's placenta. Uterine NK cells are the most abundant immune cell in the early decidua (up to 70% of decidual leukocytes in the first trimester). Rather than killing, they remodel spiral arteries and secrete angiogenic factors; certain maternal KIR / fetal HLA-C combinations are linked to pre-eclampsia and recurrent miscarriage.
• A living cancer drug. Allogeneic NK-cell infusions and CAR-NK cells are in clinical trials; a landmark 2020 trial of anti-CD19 CAR-NK cells (Liu et al., NEJM) produced responses in lymphoma with far less cytokine-release syndrome and neurotoxicity than CAR-T cells, and — because they are short-lived — an off-the-shelf, donor-derived product becomes feasible.

Common misconceptions

• "NK cells kill anything they touch." They do not. Killing is licensed only when activating signals overwhelm inhibition. Healthy cells with normal MHC class I are continuously inspected and spared — an NK cell can form a synapse with a healthy cell, sense inhibition, disengage, and move on.
• "NK cells recognize antigen like T cells do." They have no antigen-specific, somatically rearranged receptor and undergo no VDJ recombination. Their receptors are germline-encoded pattern and MHC sensors. NK cells recognize changes in the state of a cell (loss of MHC, gain of stress ligands, antibody coating), not a particular peptide sequence.
• "Missing self is the whole story." Loss of MHC class I removes the brake, but an NK cell usually still needs a positive activating signal ("induced self" — stress ligands like MICA/MICB via NKG2D) to kill efficiently. A cell that lost MHC but expresses no activating ligands may be ignored. Killing is a two-signal decision, not a single switch.
• "NK cells are strictly innate and have no memory." Outdated. Mouse Ly49H+ NK cells form long-lived, antigen-specific memory to cytomegalovirus, and human CMV drives durable "adaptive" NKG2C+ CD57+ NK expansions with epigenetic reprogramming. Cytokine-induced trained immunity also leaves NK cells primed for weeks.
• "NK cells and CTLs kill by different mechanisms." The terminal weapon is the same — directed exocytosis of perforin and granzymes plus death-receptor ligands (FasL, TRAIL). What differs is target recognition, not the lethal hit. Perforin-deficient patients (familial hemophagocytic lymphohistiocytosis, PRF1 mutations) show that both cell types depend on the same granule machinery.
• "NK licensing means every NK cell is fully armed." NK cells must be "educated" during development by engaging self-MHC through inhibitory receptors to become fully responsive — a process called licensing or education. NK cells that never engage self-MHC are hyporesponsive, a built-in safeguard that prevents autoreactivity against MHC-low healthy tissue.

How natural killer cells work

An NK cell is a large granular lymphocyte packed with pre-formed lytic granules. It decides whether to kill by summing signals at the immunological synapse it forms with a potential target. Two receptor families compete. Activating receptors include NKG2D (which binds the stress-induced ligands MICA, MICB, and ULBP1–6 that appear on damaged, transformed, or infected cells), the natural cytotoxicity receptors NKp30, NKp44, and NKp46, and CD16 (FcγRIIIa, the Fc receptor for IgG). These signal through ITAM-bearing adaptor chains — DAP12, CD3ζ, and FcεRIγ — that recruit the kinases Syk and ZAP70, while NKG2D in humans pairs with the adaptor DAP10, whose YxxM motif recruits PI3K instead; together they push toward degranulation. Inhibitory receptors — the killer-cell immunoglobulin-like receptors (KIRs) and the NKG2A/CD94 lectin heterodimer in humans, the Ly49 family in mice — bind MHC class I on healthy cells. Their cytoplasmic ITIM motifs recruit the phosphatases SHP-1 and SHP-2, which dephosphorylate activating intermediates and abort the kill. Under normal conditions the inhibitory signal dominates, so healthy cells are spared.

The missing-self logic falls out of this arithmetic. When a virus or tumor downregulates MHC class I to escape cytotoxic T cells, the inhibitory receptors lose their ligand and fall silent. The activating signals — amplified by stress ligands the target now displays ("induced self") — cross the threshold, and the NK cell commits. This elegantly explains why NK cells preferentially destroy exactly the cells that have made themselves invisible to CTLs. It is a two-signal decision: loss of inhibition plus gain of activation.

Once committed, the NK cell delivers the lethal hit. The microtubule-organizing center reorients toward the synapse and lytic granules — modified secretory lysosomes containing perforin and a family of serine proteases called granzymes — traffic to the contact site and undergo directed exocytosis into the synaptic cleft. Perforin oligomerizes in the target's plasma membrane in a calcium-dependent manner, forming pores and enabling granzyme delivery into the cytosol (largely via endosomal uptake and perforin-facilitated release). Granzyme B, the best-characterized, cleaves and activates caspase-3 and cleaves the BH3-only protein BID to tBID, driving mitochondrial outer-membrane permeabilization and apoptosis. NK cells can additionally engage death-receptor pathways, displaying FasL and TRAIL that ligate Fas/CD95 and DR4/DR5 on the target to trigger extrinsic apoptosis. A single NK cell can kill several targets serially, detaching and reloading its granule armory between hits.

ADCC is the antibody-directed shortcut. When IgG coats a target, CD16 clusters on its Fc tails and delivers one of the strongest single activating signals an NK cell can receive — strong enough to override inhibition. This makes NK cells the cellular executioners behind therapeutic antibodies, and it is why Fc engineering (afucosylation) and the FCGR3A V158F polymorphism, which tune CD16-IgG affinity, measurably change clinical antibody efficacy.

NK cells vs cytotoxic T cells

Feature	NK cell	Cytotoxic T cell (CD8+ CTL)
Immune arm	Innate lymphocyte	Adaptive lymphocyte
Antigen receptor	None — germline receptors only	Somatically rearranged TCR (VDJ)
Prior sensitization	Not required — kills on first encounter	Required — primed by dendritic cells
Response time	Hours	Days (5–7 for priming + expansion)
Sees MHC class I as	An inhibitory "self" signal	The required presentation platform
Triggered by MHC-I loss	Yes — "missing self" activates killing	No — the cell becomes invisible
Lethal weapon	Perforin/granzyme, FasL, TRAIL	Perforin/granzyme, FasL
Antibody-directed killing	Yes — ADCC via CD16 (FcγRIIIa)	No
Memory	Limited/adaptive (e.g. CMV NKG2C+)	Robust clonal memory

Activating vs inhibitory NK receptors

Property	Activating receptors	Inhibitory receptors
Examples (human)	NKG2D, NKp30/44/46, CD16, activating KIRs, NKG2C	Inhibitory KIRs (KIR2DL, KIR3DL), NKG2A/CD94, LILRB1
Ligands	Stress ligands MICA/MICB, ULBP1–6; viral/tumor ligands; IgG-Fc	Classical MHC class I (HLA-A/B/C), HLA-E
Signaling motif	ITAM (via DAP12, CD3ζ, FcεRIγ); DAP10 YxxM → PI3K	ITIM (recruits SHP-1/SHP-2)
Net effect	Promotes degranulation and killing	Aborts activation — protects healthy cells
Signal dominance	Must exceed threshold to kill	Normally dominant on healthy cells
Triggered by	Cellular stress, transformation, infection, antibody	Normal self-MHC display
Therapeutic angle	Agonists, CAR-NK, ADCC-enhancing antibodies	Checkpoint blockade (anti-NKG2A monalizumab; anti-KIR)

Famous experiments

• Kiessling, Klein, Wigzell & Herberman (1975). Two groups — one in Stockholm, one at the U.S. NCI — independently reported that lymphocytes from unimmunized mice spontaneously lysed tumor cells in vitro without T-cell help or prior exposure. What had been a nuisance background signal in tumor-immunology assays was recognized as a distinct effector cell and named the "natural killer" cell.
• Kärre & Ljunggren, missing-self (1986). Working with MHC-class-I-deficient tumor variants (RMA-S) in mice, Klas Kärre and Hans-Gustaf Ljunggren showed NK cells preferentially killed the MHC-low cells and spared the MHC-normal parent — the mirror image of T-cell logic. The missing-self hypothesis reframed NK recognition as surveillance for the absence of self.
• Ly49 and the inhibitory receptor (1990s). Wayne Yokoyama's identification of mouse Ly49 receptors, and the parallel discovery of human KIRs by groups including Eric Long and Lorenzo Moretta, gave the missing-self model its molecular basis: germline MHC-class-I-binding receptors that deliver dominant inhibition through ITIM motifs and SHP-1.
• Imai cohort, NK activity and cancer risk (2000). An 11-year prospective study of roughly 3,600 residents of Saitama, Japan (Imai et al., Lancet) found that individuals with high peripheral-blood NK cytotoxic activity had significantly lower subsequent cancer incidence than those with low activity — direct human evidence for NK immunosurveillance.
• Anti-CD19 CAR-NK trial (2020). Liu, Rezvani and colleagues (NEJM) infused cord-blood-derived CAR-NK cells into patients with relapsed CD19+ lymphoid malignancies. Responses occurred in the majority, with strikingly little cytokine-release syndrome, neurotoxicity, or graft-versus-host disease — establishing NK cells as a promising off-the-shelf cellular therapy platform.