Dendritic cells (DCs) form close contacts with antigen-specific CD4+ and CD8+ T cells, and these interactions influence both parties: T cells are activated to proliferate while the maturation of DCs is enhanced. At later stages of the immune response, T-cell killing of DCs might help to switch off immune responses. Now, a third type of DC–lymphocyte collaboration has been revealed. Three papers in The Journal of Experimental Medicine explore human DC–natural killer (NK) cell interactions and reveal a contact-dependent cross-regulation.

NK cells are innate lymphocytes that directly kill virally infected cells and secrete cytokines that have potent antiviral effects. Recent studies, mostly in mice, indicate that DCs can potentiate NK-cell activity and, under certain conditions, NK cells have been shown to kill DCs. But the DC–NK relationship had not been examined in detail. The new studies investigated DC–NK interactions by culturing human peripheral blood NK cells with monocyte-derived DCs.

Gerosa and colleagues show that DCs enhance NK-cell activation (as measured by CD69 upregulation or cytotoxic activity), but only in the presence of inflammatory stimuli such as lipopolysaccharide or interferon (IFN)-α. In turn, activated NK cells induce DC maturation, as indicated by upregulation of the costimulatory molecule CD86 and cytokine secretion. DCs matured by activated NK cells were shown to be fully capable of stimulating naive T cells in vitro.

Piccioli et al. show that the DC:NK ratio is crucial in determining the outcome of the interaction. When DCs outnumber NK cells, DCs mature and secrete cytokines, and NK cells are activated; however, if NK cell numbers overwhelm DCs, DCs are killed by NK cells and further NK activation is prohibited. So, NK–DC encounters can both amplify and attenuate innate responses.

The third study by Ferlazzo et al. shows that, in addition to CD69 upregulation and enhanced cytotoxicity, contact with DCs promotes the expansion of NK cells. Like Piccioli et al. they show that NK cells can kill DCs. They also found, however, that mature DCs are more resistant than immature DCs to NK-cell-mediated lysis.

What then is the basis of the DC–NK interaction? All three groups show that DC activation by NK cells requires cell-to-cell contact. In addition, DC activation appears to involve tumour-necrosis-factor-α produced by DCs and IFN-γ produced by NK cells. Ferlazzo et al. used blocking antibodies to show that the NK-cell receptor NKp30 is essential for the killing of DCs; however, inhibition of NKp30 has no effect on the DC-induced expansion of NK cells. The other groups also failed to identify the DC factors that mediate NK-cell activation.

Together, these studies describe a new regulatory relationship in innate immunity. In a commentary in the same issue, Lawrence Zitvogel outlines the ways in which this two-way negative and positive cross-regulation might co-ordinate the initiation and dampening down of innate responses to infection.