Natural Killer Cells and Missing-Self Recognition
A virus infects a cell and immediately begins sabotaging it. It shuts down the machinery that displays viral fragments on MHC class I molecules to CD8+ T cells. This is a clever trick. Without MHC class I on the surface, cytotoxic T lymphocytes cannot see the infected cell at all. Yet many such stealthed cells still die within hours. Something else in the immune system is patrolling for cells that have gone quiet in exactly this way. It kills them precisely because they have gone quiet. That something is the natural killer (NK) cell. It is a lymphocyte of the innate immune system. It does not need prior antigen exposure, clonal expansion, or antigen-specific receptor rearrangement to act. NK cells solve a genuinely hard sensing problem. How do you detect the absence of a normal marker, rather than the presence of an abnormal one? Their answer was formalized as the "missing-self hypothesis" by Klas Kärre in the early 1980s. It works by constantly integrating signals from two opposing families of surface receptors. Inhibitory receptors read out "this is healthy self tissue" via MHC class I. Activating receptors read out "this cell is stressed or foreign." When the inhibitory signal drops out, the balance tips toward killing. This matters because it closes a critical gap in host defense. Many viruses (herpesviruses, poxviruses, some retroviruses) and many tumors downregulate MHC class I precisely to escape CD8+ T cells. If that were a fully effective escape route, cytotoxic immunity would have a fatal blind spot. NK cells convert MHC class I loss from a safe haven into a death sentence. This forces pathogens and tumors into a lose-lose bind. They can keep MHC I and risk T cell killing. Or they can lose it and risk NK killing. This double bind is a recurring theme in immune evolution. It is also a template for how surveillance systems can be built from combinations of simple sensors, rather than one perfect one. This article develops the receptor biology of NK cells and the molecular logic of missing-self recognition. It covers the classic experiments that established the hypothesis: bone marrow chimera rejection, beta-2-microglobulin knockouts, and "hybrid resistance." It also covers modern clinical uses in transplantation and cancer immunotherapy.
- NK cells kill target cells lacking normal MHC class I ("missing self"), complementing T cells that need MHC I to detect infection.
- Killing is set by the balance of inhibitory receptors (KIRs, NKG2A/CD94, Ly49 in mice) reading MHC-I, versus activating receptors (NKG2D, natural cytotoxicity receptors) reading stress ligands.
- Cytotoxicity uses perforin/granzyme-loaded granules and Fas ligand, the same core weapons cytotoxic T cells use, but triggered by innate pattern integration, not clonal antigen receptors.
- Kärre's missing-self hypothesis was proven by hybrid resistance experiments and confirmed with MHC class I–deficient (β2-microglobulin knockout) mice, which are selectively rejected by NK cells.
- NK cells also perform antibody-dependent cellular cytotoxicity (ADCC) via CD16/FcγRIII, linking innate killing to antibody responses.
- Clinically, missing-self logic underlies NK "alloreactivity" in haploidentical bone marrow transplant, and NK-based and checkpoint therapies exploit the activating/inhibitory balance against tumors.
Natural killer cells kill by continuously weighing inhibitory signals from self MHC class I against activating signals from stress ligands, so that losing MHC class I — not gaining a foreign antigen — is itself the trigger for lysis.
The core idea
A natural killer (NK) cell is a large granular lymphocyte, part of the innate lymphoid cell family, that recognizes and kills abnormal host cells without prior sensitization. Unlike T and B cells, NK cells do not rearrange antigen-specific receptor genes; they are generated with a fixed, germline-encoded repertoire of receptors and are ready to respond within hours of encountering a target, making them a bridge between the fast, generic responses of Innate Immunity and the slow, exquisitely specific responses of adaptive immunity described in Cells of the Immune System. The defining conceptual move in NK biology is the missing-self hypothesis. Adaptive cytotoxic T lymphocytes are activated by the presence of a specific peptide–MHC class I complex; they need MHC class I to see anything at all. NK cells work by the opposite logic: they carry inhibitory receptors that recognize MHC class I molecules broadly (not particular peptides), and engagement of these receptors delivers a "stand down" signal. Healthy cells express abundant MHC class I and are protected. Cells that lose MHC class I — because a virus has degraded it, because a tumor has downregulated it to escape T cells, or because they are foreign bone marrow cells lacking self MHC I — lose this protective signal. Loss of inhibition, combined with activating signals from stress-induced ligands, tips an NK cell into a killing program. The absence of a normal molecule, not the presence of an abnormal one, is what triggers cytotoxicity — hence "missing self." The receptors involved fall into two functional classes, encoded largely (though not exclusively) by two different gene clusters. Inhibitory receptors — killer-cell immunoglobulin-like receptors (KIRs) in humans, Ly49 receptors in mice, and the CD94/NKG2A heterodimer in both — engage classical or non-classical MHC class I and carry cytoplasmic immunoreceptor tyrosine-based inhibitory motifs (ITIMs) that recruit phosphatases to dampen activation. Activating receptors — NKG2D, the natural cytotoxicity receptors (NKp30, NKp46, NKp44), and activating KIR/Ly49 isoforms — recognize stress-induced self ligands (such as MICA/MICB and ULBPs in humans, which are induced by DNA damage, heat shock, or viral infection) and signal through ITAM-bearing adaptor proteins (DAP12, DAP10, CD3ζ, FcRγ). The net output of an NK cell at any moment is the arithmetic sum of these opposing signals — a molecular voting system, not a single on/off switch.
Mechanism — how missing-self recognition and killing actually work
Step 1 — the inhibitory checkpoint. Every nucleated cell in the body normally displays MHC class I molecules loaded with peptides sampled from its own cytoplasmic proteome, the same complexes CD8+ T cells scan for foreign peptide. NK cells possess inhibitory receptors (KIRs in humans; Ly49 family in mice; CD94/NKG2A in both) that bind these MHC class I molecules in a largely peptide-independent, allele-family-dependent manner. Each individual NK cell expresses a stochastically chosen subset of these inhibitory receptors, tuned during development in a process called "licensing" or "education": NK cells that express at least one inhibitory receptor matching a self MHC class I allele become functionally competent ("licensed"); NK cells lacking any self-specific inhibitory receptor remain hyporesponsive, preventing autoreactivity. When an inhibitory receptor engages MHC class I on a healthy target cell, its cytoplasmic ITIM domains become phosphorylated by Src-family kinases, recruiting the phosphatases SHP-1 and SHP-2, which dephosphorylate activating-pathway intermediates and shut down the killing program before it starts.
Step 2 — activating signal surveillance runs in parallel. Independent of MHC status, NK cells continuously sample the target cell surface for activating ligands. NKG2D binds stress-induced ligands (MICA, MICB, ULBP1–6 in humans) that are upregulated by DNA damage responses, heat shock, malignant transformation, or viral infection. Natural cytotoxicity receptors bind a broader, less fully characterized set of ligands including some viral proteins and altered self-glycoproteins. CD16 (FcγRIII) binds the Fc portion of IgG antibodies bound to a target cell, triggering antibody-dependent cellular cytotoxicity (ADCC) — this links NK killing to the humoral response and is described further alongside Cytokines and Chemokines: Signaling Molecules of Immunity as a point where innate effector cells act on antibody-opsonized targets.
Step 3 — the balance is computed at the immune synapse. As an NK cell contacts a candidate target, both receptor classes engage simultaneously and their signals are integrated at the level of the actin cytoskeleton and the microtubule-organizing center (MTOC). If inhibitory signaling dominates (strong, matched MHC-I engagement), the MTOC does not polarize toward the target, and no lytic machinery is deployed — the NK cell disengages and moves on, typically within minutes. If activating signals dominate — because MHC class I is reduced ("missing self") or because stress ligands are highly upregulated, or both — the MTOC and associated lytic granules polarize toward the synapse.
Step 4 — degranulation and killing. Lytic granules, which are modified lysosomes containing perforin and a family of serine proteases called granzymes (A and B are best studied), traffic along microtubules to the synapse and fuse with the plasma membrane. Perforin polymerizes in the target cell membrane, forming pores (or facilitating endosomal delivery) that allow granzymes to enter the target cytoplasm. Granzyme B cleaves and activates caspases (notably caspase-3) and cleaves BID, engaging the mitochondrial (intrinsic) apoptosis pathway; granzyme A triggers a caspase-independent form of cell death. In parallel, NK cells can engage Fas ligand (FasL) on the target's Fas (CD95) receptor, activating the extrinsic apoptotic pathway. The target cell dies by apoptosis within roughly 1–4 hours of stable conjugate formation — a defined, non-inflammatory death that avoids releasing pathogen or tumor content in a way that would trigger broader tissue damage. Critically, this entire process requires no antigen-specific priming: a naive NK cell can kill an appropriately marked target on first encounter, which is why NK responses are measurable within hours of infection, well before adaptive T and B cell responses mature.
Step 5 — cytokine production reinforces the response. Activated NK cells secrete interferon-gamma (IFN-γ) and, to a lesser extent, TNF-α. IFN-γ activates macrophages, upregulates MHC class I and II expression on nearby cells (paradoxically restoring the very signal that would normally protect them, but also priming for T cell recognition), and shapes the developing adaptive response — a foreshadowing of the broader interferon network covered in Type I Interferons and Antiviral Defense, where type I interferons (IFN-α/β) produced by infected cells and plasmacytoid dendritic cells directly activate NK cells and upregulate activating ligand expression, forming a tight feed-forward loop between the interferon system and NK surveillance.
The evidence — how missing-self was established
The puzzle: hybrid resistance. The story begins with an anomaly in bone marrow transplantation biology observed from the 1950s through the 1970s. Classical transplant immunology predicted that an F1 hybrid mouse (offspring of two inbred strains, e.g., strain A × strain B) should accept bone marrow grafts from either parental strain, because the F1 expresses both parental MHC haplotypes and should recognize parental cells as "self." Yet experiments (notably by George Snell and later George Cudkowicz and Marilyn Bennett in the 1960s–70s) showed the opposite: irradiated F1 hybrid mice actually rejected bone marrow grafts from either parent strain. This phenomenon, called "hybrid resistance," could not be explained by classical T-cell-mediated allograft rejection, since the F1 host's T cells should tolerate parental MHC as self. The rejecting cells were later shown to lack the markers of T and B cells and were radioresistant, non-thymic-dependent — pointing to a distinct lymphocyte population, later identified as NK cells.
Kärre's insight and the missing-self hypothesis (1981–1986). Klas Kärre, working with Rolf Kiessling and Hans-Gustaf Ljunggren in Stockholm, noticed a pattern across many contemporaneous studies: tumor cell lines and virus-infected cells that had lost or downregulated MHC class I expression were often more susceptible to lysis by NK cells than their MHC class I-positive parental lines, while lines selected for restored MHC class I expression became relatively resistant. In 1981, Kärre proposed the missing-self hypothesis: NK cells monitor for the absence of self MHC class I as a trigger for attack, complementing T cells, which require the presence of MHC class I (displaying foreign peptide) to attack. This elegantly explained hybrid resistance too: parental bone marrow cells express only one parental MHC haplotype, so from the perspective of an F1 hybrid NK cell licensed on both parental haplotypes, the parental graft cells are "missing" the other haplotype's inhibitory signal relative to full F1 self, and the missing (heterozygous mismatch) signal is enough to trigger rejection.
Direct experimental test: the RMA/RMA-S system. Ljunggren and Kärre performed a now-classic direct test in the mid-1980s using a mouse lymphoma cell line, RMA, and a mutant subline, RMA-S, selected for defective peptide transport (a defect later understood to lie in the TAP peptide transporter), which expresses drastically reduced surface MHC class I. When injected into syngeneic mice, RMA-S cells (low MHC-I) were rejected far more efficiently than parental RMA cells (normal MHC-I), and this rejection was NK-cell dependent — it could be abolished by NK depletion (using anti-asialo-GM1 antibody or anti-NK1.1 antibody) and did not require T cells. This provided a clean, controlled demonstration that reducing MHC class I on an otherwise identical cell increases its susceptibility to NK-mediated killing.
Genetic confirmation: β2-microglobulin knockout mice. A definitive genetic test came with mice engineered to lack β2-microglobulin, an obligate subunit of MHC class I; without it, essentially no functional MHC class I is displayed on the cell surface at all. Bix and colleagues (1991, working with Raulet's group), and complementary experiments by Höglund, Liao and others in the early 1990s, showed that bone marrow or tissue grafts from β2-microglobulin-deficient mice, when transplanted into normal (MHC class I-sufficient) recipient mice, were rejected in an NK-cell-dependent, T-cell-independent manner — direct proof that loss of MHC class I is sufficient, on its own, to convert a normal self cell into an NK target, with no need for any foreign antigen. Conversely, NK cells from β2-microglobulin-deficient mice themselves showed impaired ability to reject MHC class I-deficient targets, revealing that NK cells raised in an MHC-I-deficient environment fail to acquire fully licensed inhibitory receptor function — an early clue to what would later be formalized as NK cell "licensing" or "education" (Kim et al., and Raulet and colleagues, 2000s), showing that missing-self sensing is not merely reactive but requires a developmental calibration step against the host's own MHC class I repertoire.
Molecular identification of the receptors. Through the late 1980s and 1990s, the actual inhibitory receptors were cloned and characterized: Ly49 receptors in mice (Karlhofer, Ribaudo and Yokoyama, 1992, showed Ly49A directly inhibits NK lysis of MHC class I-bearing targets) and KIRs in humans (Moretta, Ciccone, Colonna and colleagues, early-to-mid 1990s, defined human "killer inhibitory receptors" and their HLA-C, -B specificities), giving the missing-self hypothesis a concrete molecular mechanism rather than just a phenomenological pattern. NKG2D and its stress ligands were characterized later (Bauer et al., 1999, identified MICA/MICB as NKG2D ligands), completing the two-arm model of activation and inhibition described above. Together, these experiments moved missing-self recognition from an inferred pattern (hybrid resistance, tumor line comparisons) to a directly manipulated, mechanistically dissected phenomenon (RMA/RMA-S, β2-microglobulin knockouts, cloned receptor genes) — the standard by which immunological causal claims are established.
Worked examples
Example 1 — predicting hybrid resistance from haplotype logic. Consider inbred mouse strains A (MHC haplotype H-2^a^) and B (MHC haplotype H-2^b^), and their F1 hybrid, A×B, which expresses both H-2^a^ and H-2^b^ on all nucleated cells, including bone marrow. F1 NK cells become "licensed" through inhibitory receptors specific for H-2^a^ and separately for H-2^b^ (an NK cell can express either or both types, generating a diverse, overlapping repertoire). Now irradiate an F1 mouse (destroying its own bone marrow and most lymphocytes but sparing radioresistant NK cells) and transplant bone marrow from strain A alone. Reasoning through the missing-self logic:
- Strain A marrow cells express only H-2^a^, not H-2^b^.
- F1 host NK cells that are licensed via H-2^b^-specific inhibitory receptors receive no inhibitory signal from the H-2^a^-only graft cells (they are "missing" the H-2^b^ ligand those particular NK cells are tuned to expect).
- These NK cells therefore see the graft as a missing-self target and can attack it, even though classical T cell tolerance would treat strain A antigens as fully "self" in the context of an A×B F1.
- Meanwhile, T cells in the F1 recognize strain A antigens as self (since the F1 was tolerized to both parental haplotypes during thymic development) and do not reject the graft.
Prediction: F1 hosts reject parental-strain bone marrow via NK cells alone, in a manner that is radioresistant (NK cells survive irradiation doses that kill most T/B lymphocyte precursors) and does not require T cell help. This is exactly the hybrid resistance phenotype originally observed by Snell, Cudkowicz, and Bennett, and it can be abolished experimentally by depleting NK cells (anti-NK1.1 or anti-asialo-GM1 antibody) or using NK-deficient (e.g., beige) mice, which fail to reject the graft — confirming the causal role of NK cells rather than any residual T cell activity.
Example 2 — clinical reasoning: haploidentical stem cell transplant and NK alloreactivity. A patient with acute myeloid leukemia (AML) needs a bone marrow (hematopoietic stem cell) transplant but has no fully HLA-matched sibling or unrelated donor. The clinical team considers a haploidentical transplant from the patient's parent, who shares only one HLA haplotype with the patient. Consider a simplified KIR/HLA scenario. Donor NK cells develop and are licensed against the donor's own HLA class I alleles. Suppose the donor is homozygous for the HLA-C group C1 (all their nucleated cells express C1-type ligands), so donor NK cells are licensed via inhibitory KIR2DL2/2DL3 receptors specific for C1. The recipient's leukemic (and normal) cells, inherited from the other, non-shared parental haplotype, express HLA-C group C2 instead, alongside residual C1. Reasoning:
- Donor-derived NK cells (present in the graft, or reconstituting after transplant) are licensed for and expect C1.
- If leukemic blast cells in the recipient express predominantly C2 (missing the C1 ligand the donor NK cells are calibrated to see), those donor NK cells receive insufficient inhibitory signal from the leukemic cells.
- Combined with activating ligands often upregulated on leukemic blasts (e.g., NKG2D ligands such as MICA/MICB, frequently elevated by cellular stress and DNA damage in transformed cells), the balance tips toward NK-mediated killing of the leukemia — a beneficial "missing-self, alloreactive" graft-versus-leukemia effect.
- Because this recognition operates via germline-encoded innate receptors rather than antigen-specific TCR/graft-versus-host reactivity, it can, in favorable KIR-ligand mismatch combinations, reduce leukemic relapse without necessarily worsening graft-versus-host disease (GVHD), a finding first reported clinically by Ruggeri, Velardi and colleagues (Perugia group, 2002) in haploidentical transplants — a direct clinical translation of the missing-self principle discovered in mouse hybrid-resistance experiments two decades earlier.
Quantitatively, in Ruggeri et al.'s landmark 2002 study of haploidentical transplants for AML, patients transplanted from a KIR-ligand-mismatched donor (donor NK cells alloreactive against recipient, missing-self scenario as above) had markedly lower relapse rates (roughly 0% versus ~75% at 5 years in the high-risk AML subgroup analyzed) compared to matched or non-alloreactive donor combinations, and no increase in acute GVHD — because donor T cells, not NK cells, are the classical drivers of GVHD, and this graft protocol was T-cell depleted.
Common misconceptions & traps
| Misconception | Why it's wrong |
|---|---|
| "NK cells only kill cells that display something foreign or abnormal." | Missing-self killing is triggered by the absence of a normal signal (MHC-I), not necessarily the presence of anything foreign — a subtractive rather than additive detection logic. |
| "NK cells are just non-specific killers with no receptors." | NK cells express a defined, genetically encoded repertoire of both inhibitory and activating receptors; specificity comes from receptor combinations and signal integration, not antigen-receptor rearrangement. |
| "Any cell with low MHC-I is automatically killed by any NK cell." | Killing requires the net balance of activating signals to exceed inhibitory signals for that individual, licensed NK cell; unlicensed NK cells (lacking a self-specific inhibitory receptor) remain hyporesponsive even to MHC-I-low targets. |
| "NK cells are part of adaptive immunity because they show 'memory' in some contexts." | Classical NK cells are innate lymphoid cells without antigen-receptor gene rearrangement; some memory-like NK behavior exists (e.g., after cytomegalovirus infection) but it is mechanistically distinct from clonal adaptive memory. |
| "MHC class I upregulation is always protective against all immune attack." | High MHC-I protects against NK cells (via inhibitory receptors) but is exactly what makes a cell visible and vulnerable to CD8+ T cells — the two systems have opposite MHC-I dependencies. |
The scientist behind it
Origins · 1 minKlas Kärre and how this was discovered
Klas Kärre trained as an immunologist at the Karolinska Institute in Stockholm, working alongside Rolf Kiessling, Hans-Gustaf Ljunggren, and others in a research environment deeply engaged with tumor immunology and NK cell biology through the late 1970s and 1980s. Kärre's central contribution was not a single decisive experiment but a unifying conceptual synthesis: he recognized that a scattered set of observations — tumor cell lines varying in MHC class I expression and NK susceptibility, and the decades-old puzzle of hybrid resistance in bone marrow transplantation — were all instances of one underlying rule. He proposed, in his 1981 doctoral thesis and subsequent papers with Ljunggren (notably a influential 1985 and 1986 series of papers), that NK cells are tuned to detect the loss of self MHC class I molecules as a positive trigger for cytotoxicity, complementing rather than duplicating the surveillance done by CD8+ T cells. This hypothesis was then rigorously tested using the RMA/RMA-S cell line system Kärre and Ljunggren developed, and was subsequently validated genetically by other groups using MHC class I-deficient mice. Kärre's framework reoriented an entire subfield: rather than asking "what foreign thing does this receptor detect," immunologists began asking "what self signal is this cell failing to display," a question that now extends well beyond NK biology into broader innate sensing concepts.
Where this matters today
Hematopoietic stem cell transplantation. As worked through above, KIR-ligand mismatch between donor NK cells and recipient tissue is now an active consideration in donor selection for haploidentical and unrelated-donor transplants, particularly for acute myeloid leukemia, where missing-self-driven graft-versus-leukemia effects from alloreactive donor NK cells can reduce relapse.
Cancer immunotherapy. Because many tumors downregulate MHC class I to escape T cells (exactly the vulnerability missing-self recognition exploits), NK cells are an active target for therapy: engineered CAR-NK cells, monoclonal antibodies that block inhibitory KIR or NKG2A signaling (e.g., anti-KIR antibodies, anti-NKG2A antibodies such as monalizumab), and cytokine-based NK expansion protocols (using IL-2, IL-15) are in clinical trials to unleash NK-mediated tumor killing, often in combination with T-cell checkpoint inhibitors, since the two systems act on complementary vulnerabilities of the tumor.
Antibody-based therapies and ADCC. Many therapeutic monoclonal antibodies used in oncology (e.g., rituximab against CD20, trastuzumab against HER2) work partly by opsonizing tumor cells for NK-mediated antibody-dependent cellular cytotoxicity via CD16, making NK cell number and function a relevant biomarker for treatment response.
Viral infection and immunodeficiency. Rare human patients with selective NK cell deficiencies suffer severe, recurrent infections with herpesviruses (HSV, VZV, CMV, EBV) — a natural experiment confirming that NK cells provide non-redundant antiviral defense specifically against pathogens skilled at MHC class I downregulation, reinforcing the mechanistic logic developed above and setting up the broader antiviral interferon network discussed under Type I Interferons and Antiviral Defense.
Pregnancy immunology. Uterine NK cells, a distinct tissue-resident population, interact with trophoblast cells that express unusual, reduced sets of MHC class I molecules (notably HLA-G and HLA-C but not classical HLA-A/B); missing-self-type receptor interactions here are thought to regulate trophoblast invasion and placental vascular remodeling rather than triggering cytotoxicity, illustrating that the same receptor logic can be repurposed for a non-lytic, tissue-remodeling function.
Ask the tutor: How does NK cell licensing get re-calibrated in bone marrow transplant patients who now carry a new, donor-derived immune system with a different MHC class I repertoire than their original tissues?
Are NK cells part of the innate or adaptive immune system?
NK cells are classically part of the innate immune system: they use germline-encoded receptors, respond within hours without prior antigen exposure, and do not undergo the somatic gene rearrangement (V(D)J recombination) that generates T and B cell antigen receptor diversity. However, this boundary is not perfectly rigid — some NK cells, particularly a subset expanded after cytomegalovirus infection, display memory-like features such as clonal-like expansion and enhanced recall responses upon re-exposure. This "adaptive NK cell" phenomenon is mechanistically distinct from classical immunological memory (it does not involve antigen receptor rearrangement) but shows that innate and adaptive immunity exist on a spectrum rather than as two hermetically sealed categories.
What exactly is "licensing" and why does it prevent autoimmunity?
Licensing (also called NK cell education) is the developmental process by which an NK cell becomes functionally competent by engaging at least one inhibitory receptor with a matching self MHC class I allele during maturation in the bone marrow. NK cells that fail to express any inhibitory receptor specific for the host's own MHC-I alleles remain hyporesponsive — they can still express activating receptors, but their overall killing threshold is set very high, so they do not attack healthy self tissue even though they technically lack inhibitory engagement in a given encounter. This is essential because self MHC class I alleles vary between individuals; without licensing, an NK cell with no receptor for any of the host's particular MHC-I alleles could otherwise treat all healthy self tissue as "missing self" and attack indiscriminately.
How is NK-mediated killing different mechanistically from CD8+ T cell killing, if both use perforin and granzymes?
The killing machinery itself — perforin pore formation, granzyme delivery, caspase activation, FasL-Fas engagement — is essentially shared between NK cells and cytotoxic CD8+ T cells, since both are specialized lytic effector lymphocytes. The difference lies upstream, in how the decision to kill is made. CD8+ T cells require clonal antigen-receptor engagement with a specific peptide-MHC class I complex, following prior priming and clonal expansion, so their targeting is antigen-specific and adaptive. NK cells make the decision by integrating germline-encoded inhibitory and activating receptor signals without needing a specific antigen, so their targeting is based on overall cell "health signature" (MHC-I level plus stress ligand expression) rather than a single specific epitope.
Can viruses evolve to evade both T cell and NK cell surveillance simultaneously?
Some highly adapted viruses, particularly cytomegalovirus (CMV), have evolved sophisticated dual-evasion strategies. CMV encodes proteins that downregulate classical MHC class I (evading CD8+ T cells) while simultaneously expressing MHC class I-like decoy molecules (such as the mouse CMV protein m157, which engages inhibitory Ly49 receptors) that substitute for the lost inhibitory signal, preventing missing-self-triggered NK attack. This arms race is itself strong evidence for the biological importance of missing-self recognition — if it were not a genuinely dangerous threat to the virus, there would be no selective pressure favoring such elaborate decoy evolution. Host genomes counter-evolve too: some Ly49 receptor variants have become activating rather than inhibitory specifically to recognize m157 as a "danger" signal, restoring NK detection.
Why don't NK cells attack all cells that merely have somewhat reduced MHC-I, such as red blood cells?
Mature red blood cells lack nuclei and essentially all MHC class I entirely, yet they are not attacked by NK cells, which seems to contradict a naive reading of missing-self. The resolution is that NK cell activation requires a positive activating signal in addition to loss of inhibition; red blood cells present essentially no activating ligands (no NKG2D ligands, no stress markers) because they are terminally differentiated, enucleated, and metabolically minimal, so there is nothing for activating receptors to engage. Missing-self is necessary but not sufficient for killing — the target must also carry sufficient activating ligand density, which is why the rule is best stated as a balance, not a single MHC-I threshold.
How does missing-self recognition relate to organ transplant rejection more broadly?
Classical transplant rejection is dominated by T cells recognizing foreign MHC molecules directly (or foreign peptides presented on host MHC) as antigen, which is a "presence of foreign signal" mechanism distinct from missing-self. NK cells contribute a separate layer specifically relevant when MHC class I is mismatched or absent, as in hybrid resistance or haploidentical bone marrow transplantation, where donor or recipient cells may lack the specific inhibitory ligand a given NK cell is licensed against. In solid organ transplantation, NK cells are a secondary but increasingly recognized contributor to both acute and chronic rejection and, in some contexts, to graft tolerance, depending on the precise donor-recipient KIR/HLA combination.