Those immunologists old enough to remember doing experiments at a time when music by the Beatles might have been playing in the background, data were plotted with pencils and graph paper, and gene cloners were the bacteriologists somewhere down the corridor, will remember that T cells were either OKT4 or OKT8, the latter population confusingly termed ‘suppressor/cytotoxic’.1 Those for whom all the above terms are mysterious should find an ageing laboratory head to ask. Nowadays, OKT8 cells are termed CD8, and mechanisms by which they can regulate or suppress other responses have been characterized in a number of examples.2,3 One mechanism by which this can occur is through the release of cytokines that condition dendritic cells (DC), leading to immune deviation of DC-stimulated responses. A number of recent papers, including the work from Noble and colleagues in this volume, characterize the ability of various T-cell populations to modulate DC function in this way.4–8 DC are pivotal antigen-presenting cells (APC) in the immune system, because of their unique ability to induce primary immune responses.9 There exists a complex, bi-directional interaction between the DC and the T cell that determines the nature of the response that the activated T cell will subsequently mount. Viewed as a conventional interaction between an APC and a CD4 T cell, polarization of the T cell to a T helper type 1 (Th1) or Th2 phenotype may be determined by several factors, including the nature, dose and binding affinity of the antigenic peptide, structural features of the T-cell receptor (TCR), and the expression of co-stimulatory molecules.10 Furthermore, the nature of the DC may itself confer inherent skewing to a Th1 or Th2 outcome. Two publications during 1999 defined DC subpopulations, in mice and humans, respectively, that conferred Th1/Th2 skewing on the subsequent responses.11,12 Rissoan and colleagues, working with human DC, showed that peripheral-blood-monocyte-derived, interleukin-4 (IL-4)/granulocyte–macrophage colony-stimulating-factor-expanded immature myeloid DC (DC1) preferentially induce a strong interferon-γ (IFN-γ) response from co-cultured, allogeneic CD45RO− CD4 T cells.11 Conversely, the DC2 population, derived from plasmacytoid cells preferentially induce IL-4, IL-5 and IL-10 from T cells. While the precise developmental relationship between the subsets of DC that have been characterized in mice and humans is not yet clear, similar observations were made by Moser's group working in a murine model.12 CD8α+ or CD8α− DC were pulsed with keyhole limpet haemocyanin and used to prime mice in vivo. Priming with the former population led to a polarized Th1 response and priming with the latter population gave a Th2 response. In the context of a dynamic immune system with APC presenting diverse antigens to T cells, it may be inappropriate to hard-wire the T-cell cytokine response in a fixed way to the identity of the presenting DC. Indeed, it now appears that there is a high degree of flexibility and even reversibility with respect to modulation of DC function by local cytokines or stimulation with pathogens.13,14 If DC require cytokine priming to activate T cells, but T cells are the main source of immune cytokines, how can the arming process ever proceed? The answer of course is that in the physiological context, this cytokine milieu for DC maturation and polarization can be triggered as a direct result of DC exposure to infectious agents, that is, through innate immunity.14,15 Binding of bacterial lipopolysaccharide to toll-like receptor (TLR)-4 on DC leads to activation for Th1 responses.15 DC can also be polarized to stimulate Th1 responses by Bordetella pertussis toxin, while different strains of Candida albicans can license for either Th1 or Th2 activation (Fig. 1).16,17 In general, the licensing of DC for Th1 activation is associated with the stimulation of tumour necrosis factor-α and/or IL-12 from the DC. In studies using cholera toxin, heat-killed yeast and zymosan granules, all of which have been shown to promote the activation of Th2 responses by DC, this can ensue from the induction of IL-10, inhibiting the release of Th1-promoting cytokines, although it is not invariably accompanied by IL-10 release.14,18 Figure 1 Mediators and cell types known to license dendritic cells for T helper type 1 (Th1) and Th2 activation. Recently another facet has been added to this model with the realization that the licensing of DC for activation of T cells may be achieved not just through innate recognition of infectious agents but through the release of cytokines from T cells. That is, T cells activate or differentiate DC permitting appropriate presentation to T cells. This presupposes that there are populations of T cells with the ability to mount a rapid cytokine response at the very initiation of presentation. A number of recent papers address the nature of these T cells and the impact of their DC programming. Indeed, even in the context of activation by bacterial products such as lipopolysaccharide, DC can lose the ability to produce IL-12 on subsequent CD40 ligation by naive cells unless supplied with T-cell-derived IFN-γ.19 One solution to the problem of a rapid T-cell response to arm DC for Th1 responses comes from the observation that immature, human, myeloid DC express very low levels of human leucocyte antigen class II, but abundant CD1 molecules which can be recognized by the γδ receptors of Vδ1 cells.5 This maturation process involves the production of both tumour necrosis factor-α and IL-12. In another example, using lymphochoriomeningitis virus TCR transgenic T cells, naive CD8 T cells were able to support DC maturation.4 A related model used alloreactive, naive CD40L-negative CD8 cells that were able to respond to class I/peptide with an extremely rapid, DC-polarizing IFN-γ response.6 The experiments reported by Noble and colleagues in this issue add to this story and reiterate the DC regulatory role of CD8 cells.8 The groups of Noble and Kemeny have proposed that CD8 cells can regulate immunoglobulin E (IgE) responses; the present findings offer a clear mechanism for this through the CD8-mediated modulation of DC function for immune deviation from a Th2 to a Th1 phenotype.5 In an earlier paper using CD8 cells from the OT-I TCR transgenic line specific for ovalbumin it had been shown that CD8 T cells had the ability to inhibit the IgE response to antigen. This was not dependent on a direct effect of T-cell-derived IFN-γ on Th1/Th2 skewing, but on the ability of CD8 cells (but not CD4) cells to induce IL-12 release from DC.5 In the present paper, this principle is extended to the analysis of the graft-versus-host reaction, which also encompasses hyper-IgE reactivity. The experimental model is the transfer of parental strain splenocytes into (BALB/c × C57BL/6)F1 recipients. The control of Th1/Th2 responses and IgE class switching by CD8 cells is confirmed in this model since the removal of CD8 cells from the donor population causes a dramatic increase in IgE production. There is also an increase in the IgG1 anti-DNA autoantibody response. Again, the effect of the CD8 population appears to be independent of their own IFN-γ release but dependent on their ability to elicit a rapid IL-12/IL-18 response in the DC. A better understanding of these events should open up another front on which we can hope to manipulate the Th1/Th2 polarization that contributes to inappropriate immune reactivity in autoimmunity and asthma. The knowledge that both innate recognition of pathogens by DC and specific activation of CD8 T cells can create the conditions for reprogramming of DC function and subsequent CD4 T-cell immune deviation offers a possible solution to the conundrum of the diverse array of infectious agents that can be causally related to the onset of autoimmunity. In the meantime, there is much work to be done on the detailed characterization of the very early events at the top of this cascade. We need to know which are the populations of CD8 cells that mediate this DC arming in vivo, precisely which cytokines are involved and what form of activation and signalling are required. Indeed, we lack even a clear model for what could be the nature of the DC maturational differences that would preferentially skew responses to Th1 or Th2. For those too deferential to confront their ageing laboratory head, OKT4 and OKT8 were the early antibodies for human CD4 and CD8 cells and the Beatles were a pop group.