Integrated immune responses to infection – cross‐talk between human γδ T cells and dendritic cells

γδ T cells have been conserved for at least 400 million years and are the prototype of unconventional lymphocytes, displaying characteristics of both T cells and natural killer (NK) cells.1 Although generally portrayed as a minor subset of human T cells, peripheral blood γδ T cells rapidly proliferate following infection by certain pathogens, sometimes expanding from 1% to > 50% of circulating T cells within a week. The means by which this expansion occurs and its biological consequences are thus pivotal for an understanding of human responses to infection. In this issue of Immunology, Kunzmann and colleagues provide useful insight into how the actions of γδ T cells, dendritic cells (DC) and NK cells may be co-ordinated at early time points after infection.2 Moreover, the results may be relevant to tumour immunology and to the growing application of γδ T cells in clinical immunotherapy. Since the initial discovery by Pfeffer and colleagues that the major peripheral blood γδ T-cell population, the Vγ9/Vδ2 subset, is activated by low molecular weight, proteinase K-resistant compounds in a major histocompatibility complex (MHC)-independent manner,3 a plethora of ‘non-peptidic’ or ‘phosphoantigens’ has been described.4–6 These include phosphorylated and pyrophosphorylated natural metabolites such as 2,3-bisphosphoglycerate and isopentenyl pyrophosphate (IPP); synthetic molecules, such as monoethyl phosphate and IPP derivatives such as bromohydrin pyrophosphate; ubiquitous alkylated (and non-phosphorylated) amines such as isobutylamine and ethylamine; and phosphonated compounds like the aminobisphosphonate drugs pamidronate, alendronate and zoledronate (Fig. 1). More recently it has become evident that many of these molecules of intermediate and low bioactivity (and possibly little physiological significance) may act by mimicking the highly potent, natural Vγ9/Vδ2 T-cell activator (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMB-PP; EC50 0·1 nm).12 HMB-PP is an intermediate of the non-mevalonate pathway of IPP biosynthesis that is present in all Vγ9/Vδ2 T-cell-stimulating microbial pathogens (e.g. Escherichia coli, mycobacteria and malaria parasites), but is apparently absent from non-stimulating micro-organisms (e.g. streptococci, staphylococci and fungi), as well as from the human host.13 Thus, the capacity of Vγ9/Vδ2 T cells to be activated by HMB-PP facilitates a specific response to numerous microbial pathogens, by targeting a distinctive and vital metabolic route shared by these organisms. Bioactivities of nonpeptidic activators, as assembled from the literature (references 6–11). This rapid HMB-PP-driven response to a broad range of pathogens evokes aspects of innate immunity. Nonetheless, the mechanics of this recognition remain enigmatic. Although the presence of a functional Vγ9/Vδ2 T-cell receptor (TCR) is mandatory for the reaction to non-peptidic activators, and a transfected TCR will confer responsiveness on a hitherto unresponsive cell,14 there is as yet no evidence for direct interaction with the TCR, even from studies that attempted co-crystallization of the TCR with non-peptidic activators.15 Moreover, it remains a challenge to envision how the binding of such small molecules to the TCR might be sufficient to trigger T-cell activation.16 Until this dilemma is resolved it is not possible to assess whether HMB-PP is recognized via specific high-affinity antigen-combining sites, generated by somatic gene rearrangement – the hallmark of any adaptive response. Owing to the lack of a rodent model for the Vγ9/Vδ2 T-cell response, relatively little is known about how the antigenic stimulation of Vγ9/Vδ2 T cells occurs in vivo. In particular, we remain uncertain whether there is a critical role for professional antigen-presenting cells (APC), akin to the peptide–antigen presentation to conventional αβ T cells by Toll-like receptor (TLR)-activated DC. While monocytes (MO), macrophages, and DC are not required for Vγ9/Vδ2 T-cell expansion driven by soluble activators in vitro, the response is substantially enhanced by cell–cell contacts.17,18 Moreover, in experiments using whole mycobacteria, antigen uptake and processing is required to stimulate the Vγ9/Vδ2 T-cell response, with the relevant activating molecules remaining stably associated with the APC surface.19 Likewise, data from pulsed target cells suggest that human Vγ9/Vδ2 T cells may recognize aminobisphosphonates bound to the surface of MO-lineage cells rather than as free forms.7,20 Recently, Favier et al. showed that the Vγ9/Vδ2 TCR of cells primed with soluble non-peptidic activators establishes with the MO line, THP-1, a mature immunological synapse enriched in phosphotyrosine, CD2, CD94 and NKG2D, yet excluding CD45 and CD69.21 Nonetheless, none of these studies have clarified whether the accessory cells actually present HMB-PP (and its analogues) via an unknown mechanism – perhaps resembling the presentation of CD1d-bound lipid antigens to Vα14 NKT cells22– or whether they activate the cells via co-stimulation, integrins, or cytokines. The means by which γδ-cells become activated in vivo is all the more enigmatic in the case of virus infections and tumour immunity. Although some viruses, such as Epstein–Barr virus, and tumour cell lines such as Daudi Burkitt or Molt-4 lymphoma provoke striking expansion of Vγ9/Vδ2 T cells,23,24 the ‘antigens’ responsible are not known.25 While it has been hypothesized that stressed cells may over-express the low bioactivity agonist IPP,26 the response to stressed cells need not necessarily involve the TCR, because other γδ T-cell subsets, e.g. Vγ1/Vδ1+ cells, show similar recognition patterns.27 The idea of a TCR-independent response of γδ T cells to virus infection is bolstered by the finding that NK cells, with which γδ T cells are commonly compared,28 express both TLR3, a cellular receptor for double-stranded RNA (dsRNA) that is produced during myriad virus infections, and the Toll/interleukin-1 receptor domain (TIR)-containing adaptor, that transduces TLR3 signals. Consequently, dsRNA can strongly and directly activate NK cells,29 and the same might hold true for γδ T cells. Alternatively, γδ T cells might not be able to respond directly to dsRNA via TLR3, but instead respond indirectly to type I interferons (IFN) produced by DC that themselves are alerted by dsRNA. In this regard, the unconventional γδ T cells would be more closely related to conventional T cells than to NK cells, since CD8+αβ T cells are also strongly, albeit partially, responsive to type I IFN produced by DC.30,31 Indeed, it is the latter model that is supported by the results presented on page 369 of this issue of Immunology.2 The studies show that dsRNA, mimicked by poly(I:C), exerts a clear yet partial activation of peripheral blood Vγ9/Vδ2 T cells. The activation marker CD69 is up-regulated, but both CD25 and human leukocyte antigen DR (HLA-DR), that would reflect full activation, are unaffected. Likewise, there is no induction of IFN-γ production. Conversely, the up-regulation of all these parameters is considerably greater when the cells are stimulated with poly(I:C) in the presence of IPP, rather than by either poly(I:C) or IPP alone. This co-stimulation is abrogated by depletion from the cultures of CD11c+ immature DC (iDC), whereas it can be fully reconstituted by supernatants of TLR-activated iDC. By use of monoclonal antibodies, Kunzmann and colleagues were able to show that the active component in the supernatants of iDC was type I IFN. As with conventional CD8+αβ T cells, the indirect production of type I IFN is insufficient to fully activate Vγ9/Vδ2 T cells, but rather co-stimulates them, thereby substantially enhancing signals received through the TCR. As the authors acknowledge, their results are reminiscent of earlier findings by Rothenfusser et al. who observed a similar co-stimulation of γδ T cells by DNA containing unmethylated CpG motifs that provoked TLR9-mediated type I IFN production by plasmacytoid DC (pDC).32 The likely role for DC and TLRs in co-stimulating human γδ T cells is corroborated by the recent demonstration that blood-stage schizonts of Plasmodium falciparum promote γδ T-cell proliferation and IFN-γ secretion via the TLR9-MyD88 driven expression of type I IFN by pDC.33 Rupture of blood-stage schizonts would thereby release soluble molecules triggering TLRs on pDC, and at the same time γδ T cells would recognize the parasite-derived HMB-PP. Since pro-inflammatory cytokines released by γδ T cells can induce DC maturation as evidenced by up-regulation of HLA-DR, CD83 and CD86, and production of bioactive interleukin-12 (IL-12; p70),34 this self-sustaining loop would solely depend on the two unique danger signals ‘free nucleic acid’ and antigen, in this case HMB-PP. Thus, one can hypothesize how early events following infection may unfold. TLR ligands such as dsRNA and CpG-containing DNA will trigger MO-lineage cells, including iDC, to become activated APC, migrating to sites of interaction with naïve conventional αβ T cells that, over the course of the following days, will mount the adaptive cellular response. In the meantime, NK cells will be activated directly via TLRs, overseeing an early phase of immuno-protective cytolysis and cytokine release, and additional activation of further lymphoid cells. In parallel, the type I IFN generated by activated DC will co-stimulate any γδ T cells that are engaged by pathogen-derived HMB-PP, or perhaps, by stress markers presented on infected cells. Those cells can be expected to deliver potent immuno-protective responses, including cytolysis and IFN-γ and tumour necrosis factor-α (TNF-α) production: the fact that Vγ9/Vδ2 T cells can encounter iDC in the blood, while likewise, other γδ T-cell subsets and other iDC may interact in the tissues, implies that γδ T-cell activation may provide a first-line response ‘at the scene of the crime’; conversely, activation of the conventional αβ T-cell response is delayed by the routing of information flow through the lymph nodes and spleen. This might also be consistent with the idea that antigen recognition by γδ T cells may be restricted by MHC Class IB antigens such as CD1 that are expressed by iDC in the tissues.35,36 Were up-regulated IPP to be the relevant antigen for the responding γδ T cells in virus infection and tumour defence, the co-stimulation by type I IFN might facilitate the recognition of a ‘cross-reactive antigen’ (the true antigen being HMB-PP), that would in vivo be too weak to provoke a significant immune response. Alternatively, an explanation that is not ruled out by Kunzmann and colleagues is that the IFN from iDC might co-activate γδ T cells receiving a stimulus via an activating NK receptor such as NKG2D, rather than via the TCR. NKG2D engages MHC class IB molecules such as MICA and ULBP (human), and Rae1 and H60 (mouse) that are expressed by virus-infected or tumour cells.37,38 Whereas conventional T cells activated via the TCR use the adaptor DAP10 to deliver a co-stimulatory signal 2 from NKG2D, NK cells can derive both signal 1 and signal 2 from NKG2D alone since they express an additional adaptor, DAP12, that delivers signal 1 from a variant splice form-encoded product NKG2DS.39 As it was recently shown in the mouse that unconventional T cells, including γδ T cells, also express DAP12 and NKG2DS,28 the possibility remains that the γδ T-cell response to virus infection and transformed tumour cells is non-antigen-specific, i.e. TCR-independent. In addition to providing an early immuno-protective barrier to virus infection, as Kunzmann and colleagues consider,2 the γδ T-cell response in the blood and/or the tissues can also be expected to regulate the ensuing adaptive response. Indeed, in mice infected with Listeria monocytogenes, NK cell production of IFN-γ is up-regulated via the action of macrophage-derived TNF-α that is in turn up-regulated by IFN-γ produced by activated γδ T cells.40,41 Thus, one can begin to make testable predictions as to how DC, NK cells and γδ T cells may interact with each other early after infection, and how these interactions will affect both primary control of the pathogen, and the scope and shape of the subsequent αβ T-cell response. In this regard, one wonders whether differences in the response of the MO/DC compartment to infection are able to skew the γδ T-cell response, in an analogous fashion to their skewing of the αβ T-cell response. It is known that Vγ9/Vδ2 T cells can be readily coactivated by many cytokines including IL-2, IL-12, IL-15 and IL-21, and moreover, that the cells respond differently to different cytokines.42,43 While the predominant effector capability of γδ T cells in humans and mice is of a cytolytic/T helper type 1 (Th1)-type,44 it is clear that γδ T cells can exert Th2 functions.45–47 Indeed, Ferrick and colleagues showed nearly a decade ago that mouse γδ T cells are able to discriminate early in infection between bacterial and helminth pathogens by producing either IFN-γ or IL-4.48 By establishing the potential of γδ T cells to respond to cytokines made by activated DC, Kunzmann et al. open up ideas as to how such changes in the γδ T-cell response may occur. Fitting together the way in which different cellular interactions compose an integrated immune response is an essential step in immunology. As Kunzmann and colleagues suggest in this issue of Immunology, it also provides important clues as to how to manage specific components of the response, in the laboratory, and in the clinic. A.H. thanks the Wellcome Trust, and M.E. and H.J. thank the European Commission (ICA4-CT-2001–10078) and the Else Kröner-Fresenius-Stiftung for support.