Natural killer cells are versatile white blood cells that act in the innate immune system. Quite how adaptable they can be in the absence of other, more specialized, immune cells comes as a surprise.
Contact dermatitis is an occupational hazard for many people. This disruptive immune reaction can be caused by contact with an odd assortment of items, including plants, drugs, chemicals and even money1. In each case, small, reactive molecules from the item penetrate the skin and modify proteins within the tissue. Thus altered, the skin proteins are seen as foreign by the immune system, which then endeavours to destroy them. This dermatitis is caused by an adaptive response2: the reaction occurs only if that specific antigen has been encountered before, and it is mediated by the archetypal white blood cells, or lymphocytes, of adaptive immunity — activated antigen-specific αβ T cells. In contact sensitivity, the mouse experimental model of contact dermatitis, three other types of lymphocyte are implicated in the reaction: B-1 B cells, NK T cells and γδ T cells2. These cells are part of the fast-acting defences of innate immunity, which do not strengthen with successive exposures to the antigen. Notably absent from this list is the natural killer cell. So the report by O'Leary et al. in Nature Immunology3 is unexpected. They find that in mice lacking all other lymphocytes, natural killer cells play all the roles necessary to produce contact sensitivity.
Natural killer cells are different from other lymphocytes because during cell maturation they do not need to rearrange the genes for their antigen receptors (T-cell receptors or immunoglobulins). In mutant mice lacking the Rag2 gene (Rag2− mice), such gene rearrangement is blocked, so B cells, T cells and NK T cells do not exist, and natural killer cells are the only circulating lymphocytes. Unexpectedly, contact-sensitivity reactions against the chemical dinitrofluorobenzene (DNFB) (Fig. 1) are almost as strong in Rag2− mice as in normal mice3. In exploring this observation, O'Leary et al. show that, in the absence of other lymphocytes, natural killer cells are necessary for the contact-sensitivity reaction. They also find the reaction to be antigen-specific, so that Rag2− mice primed with DNFB respond to further contact with DNFB but not to a first contact with a different reaction chemical.
Not too long ago, natural killer cells were considered to be all exactly the same and lacking any antigen specificity. But O'Leary et al. add to a body of evidence showing how specific subpopulations of natural killer cells are activated in response to bone-marrow transplants4,5, viral infections6, tumours7 and, now, as part of contact-sensitivity reactions to chemically modified skin tissue3. Natural-killer-cell subpopulations are distinguished by their differential expression of a variety of activating and inhibitory receptors, of which the Ly49 receptors are notably diverse and variable among individuals8. Contact sensitivity to DNFB is mediated by the subpopulation of natural killer cells expressing Thy1, an activation marker, and either the Ly49C or Ly49I receptor. These cells concentrate in the liver3, where DNFB is probably detoxified.
The experiments of O'Leary et al. touch upon immunological memory, the property that has most clearly distinguished innate from adaptive immunity. Surviving a childhood infection with mumps or measles virus provides immunity from the disease for the rest of a person's life. In current thinking, this protection is provided by ‘memory’ B and T cells, as part of adaptive immunity, that emerge during the immune response to the infection and thereafter ensure a quick, strong and overwhelming response to further infection.
After priming Rag2− mice with DNFB, O'Leary et al. show that a secondary challenge, made either 5 or 28 days later, gave a strong response (Fig. 1). Their interpretation, that priming resulted in a persistent memory mediated by natural killer cells, is provocative and should stimulate investigation. For example, the difference between 5 and 28 days is not a particularly demanding test of persistent memory, even for a laboratory mouse with a lifespan of around two years. And, in O'Leary and colleagues' experimental design, it is possible that the primary response to DNFB had not truly subsided by the time of the secondary challenge. Elimination of DNFB and of DNFB-mediated tissue components from the mouse's painted back could take considerable time, resulting in persistent stimulation, rather than the development of memory, throughout the experiment's duration. This sort of persistent stimulation is what causes occupational contact dermatitis in humans, for instance when sensitized supermarket clerks daily handle nickel-containing coins.
In the course of evolution, lymphocytes became more diverse and more specialized. But dependency accompanies specialization, as witnessed in the Rag2− mouse — B cells, NK T cells and T cells are so reliant on their particular antigen receptors that they die when they cannot make them. By contrast, natural killer cells use many different receptors and are less dependent upon any one of them. The Ly49 receptors are a case in point, as they are flourishing in mice but do not occur in human natural killer cells9.
The study of O'Leary et al. vividly illustrates the functional versatility and adaptability of natural killer cells. In normal mice, where contact sensitivity has little natural-killer-cell contribution, the response involves at least four other lymphocyte types. In Rag2− mice, in which natural killer cells flourish in the absence of other lymphocytes, they are able to produce an equivalent contact sensitivity by themselves. A parallel can be seen in bone-marrow transplantation, where eliminating T cells from the graft allows helpful natural killer cells to emerge5. In their adaptability, versatility and deference to more specialized lymphocytes, natural killer cells seem to be the modern lymphocyte most like those lymphocyte ancestors that populated immune systems before the evolution of antigen-receptor genes that required rearrangement.