1. Immunobiology Laboratory, Cancer Research UK London Research Institute, Lincoln's Inn Fields Laboratories, 44 Lincoln's Inn Fields, London WC2A 3PX, UK
2. Microbiology and Tumor Biology Center, Karolinska Institutet, Nobelsväg 16, Solna, SE-171 77 Stockholm, Sweden
3. Department of Vaccine Research at Swedish Institute for Infectious Disease Control, Nobelsväg 18, 171 77 Stockholm, Sweden
4. Section of Immunobiology, Yale University School of Medicine and Howard Hughes Medical Institute, 330 Cedar Street, PO Box 208011, New Haven, Connecticut 06520, USA
5. These authors contributed equally to this work
6. Present address: Centre d'Immunologie de Marseille-Luminy, CNRS-INSERM, Parc Scientifique et Technologique de Luminy – Case 906, 13009 Marseille, France

Correspondence to: Caetano Reis e Sousa¹ Correspondence and requests for materials should be addressed to C.R.S. (Email: caetano@cancer.org.uk).

In mice, plasmacytoid dendritic cells (DCs) and CD8⁺ DCs appear to be the major APC subtypes in priming antiviral CTL^{2, 3, 4, 5, 6}. The role of CD8⁺ DCs in this regard might be a consequence of their unique ability to re-present cell-associated antigens on major histocompatibility complex (MHC) class I and class II^{7, 8}; this in turn might be related to their ability to phagocytose dying cells^{9, 10}. However, this process per se is insufficient to induce CTL priming because cross-presentation of cellular material by CD8⁺ DCs has also been implicated in CD8⁺ T cell tolerance^{11, 12}. Therefore, virally infected cells, but not their uninfected counterparts, must generate signals that act on DCs to favour cross-priming. One such signal might be secreted type I interferons (IFN-/)¹³. Here, we asked whether other signals are preserved in virally infected cells. Viral infection is accompanied by production of dsRNA, a potent innate stimulus that is recognized, in part, by TLR3 (ref. 14). We have previously found that CD8⁺ DCs express high amounts of messenger RNA for TLR3 but not for TLR7, a receptor for single stranded RNA prominently expressed by plasmacytoid DCs¹⁵ (see Supplementary Fig. 1a, b). Consistent with those data, exogenous synthetic dsRNA (poly I:C) activates CD8⁺ DCs but not plasmacytoid DCs; the opposite is seen with a TLR7 ligand¹⁵ (see Supplementary Fig. 1c). On the basis of these observations, we hypothesized that TLR3 allows CD8⁺ DCs to 'sense' the presence of dsRNA within virally infected cells and induce cross-priming.

We first determined whether dsRNA activates CD8⁺ DCs when supplied in cell-associated form. Purified CD8⁺ spleen DCs were co-cultured with Vero cells that had been loaded with poly I:C by electroporation (Vero-poly I:C cells) and exposed to ultraviolet light to induce cell death. Both Vero-poly I:C and mock-treated cells were phagocytosed by CD8⁺ DCs, as determined by flow cytometry and confocal microscopy (see Supplementary Fig. 2). However, compared with control cells, Vero-poly I:C cells induced de novo expression of Ifna and Ifnb genes in CD8⁺ DCs and promoted increased expression of mRNA for interleukin-6 (IL-6), tumour-necrosis factor- (TNF-), CD40 and CD86 (Fig. 1a). Higher levels of cell-surface expression of CD40, CD86 and CD80 were observed (Fig. 1b), as was the accumulation of IL-6 protein in cell culture supernatants (Fig. 1c). In contrast, IFN-, IFN- and TNF- protein levels remained below the detection limit of the enzyme-linked immunosorbent assay (ELISA), and IL-12 p40 was induced by Vero cells that had not been loaded with poly I:C (Figs 1a and 2a). The ability of poly I:C-loaded cells to activate CD8⁺ DCs was not restricted to Vero cells and could be observed using a variety of cell lines (Fig. 1c), as well as with syngeneic or allogeneic mouse splenocytes (see Supplementary Fig. 3). Ultraviolet irradiation of the target cells was not essential for CD8⁺ DC activation (see Supplementary Fig. 3), perhaps because dsRNA is a pro-apoptotic stimulus¹⁶ or because cell death is not essential for uptake¹⁷. Notably, 20–50-fold less poly I:C was required in cell-associated form than in soluble form to induce CD8⁺ DC activation (Fig. 1d). This was not due to a synergistic effect of the dsRNA and the cells (Fig. 1e), but might instead reflect differences in the uptake of particulate versus soluble dsRNA, as well as increased local concentration of the stimulus owing to settling of the poly I:C-bearing cells.

Figure 1: In vitro activation of CD8⁺ DCs by poly I:C-loaded or virally infected cells.

CD8⁺ DCs, Vero-poly I:C cells or mock-treated Vero cells were cultured in the indicated combinations. a, mRNA was analysed by RT–PCR after normalizing for HPRT. All primers except those for -actin are mouse-specific. b, Flow cytometric analysis of co-stimulatory molecule expression by CD8⁺ DCs. Vero-poly I:C cells (thick line), mock-treated Vero cells (thin line). c–g, IL-6 accumulation in the cell culture supernatants. c, CD8⁺ DCs were cultured with the indicated ultraviolet light-treated cell lines and electroporated in the presence (black bars) or absence (white bars) of poly I:C. d, CD8⁺ DCs were cultured with soluble (open circles) or Vero-associated (filled circles) poly I:C or (e) with soluble poly I:C in the presence (filled circles) or absence (open circles) of ultraviolet light-treated Vero cells. f, CD8⁺ DCs were cultured with EMCV or EMCV-infected Vero cells. Cells loaded with poly I:C were used as a positive control. g, As for (f) but using SFV. MOI, multiplicity of infection.

High resolution image and legend (50K)

Figure 2: CD8⁺ DC activation by poly I:C-loaded or virally infected cells requires endosomal recognition by TLR3.

a–c, Levels of IL-6 and IL-12 p40 in culture supernatants. a, CD8⁺ DCs were co-cultured with poly I:C- or mock-treated Vero cells in the presence of 1 M latrunculin B (black), 25 M chloroquine (grey), or no inhibitor (white). b, CD8⁺ DCs from Pkr^-/-, Myd88^-/- or wild-type (WT) control mice were cultured with poly I:C- or mock-treated Vero cells. c, As for (b) but comparing WT to Tlr3^-/- CD8⁺ DCs and also testing Vero cells infected with EMCV or SFV-OVA. CpG (0.5 g ml^-1) was used as a control stimulus.

High resolution image and legend (76K)

To determine whether virally infected cells could similarly activate CD8⁺ DCs, we first screened for viruses that do not infect CD8⁺ DCs or induce their activation directly. We found that encephalomyocarditis virus¹⁸ (EMCV) and Semliki Forest virus¹⁹ (SFV) have only a limited ability to infect CD8⁺ DCs in vitro (see Supplementary Fig. 4) and do not induce CD8⁺ DC activation when added as free viral particles (Fig. 1f, g). In contrast, both viruses infect and replicate efficiently in Vero cells (see Supplementary Fig. 4). Notably, when infected Vero cells were co-cultured with CD8⁺ DCs, they promoted DC activation as efficiently as poly I:C-loaded controls (Fig. 1f, g). This was not simply a quantitative effect because the infected cells contained less virus than the amount added in free form (data not shown). Furthermore, the SFV used for these experiments is a 'suicide' virus, engineered to undergo only one round of infection and producing no progeny virions (owing to the absence of viral structural genes)²⁰. Thus, viruses that fail to activate DCs when present as free viral particles can stimulate DCs when presented as virally infected cells.

Blocking phagocytosis using actin-stabilizing agents such as latrunculin B, or neutralizing endosomal pH with chloroquine completely prevented CD8⁺ DC activation by poly I:C-loaded Vero cells (Fig. 2a). The same drugs did not inhibit the poly I:C-independent induction of IL-12 p40 (Fig. 2a), serving as a control for the functional integrity of the cells. These results indicate that CD8⁺ DC activation is not due to factors released by dsRNA-bearing cells, and suggest instead that phagocytosis of cellular material and subsequent phagosomal acidification are necessary. To identify the pathways involved, we analysed the activation of CD8⁺ DCs lacking candidate genes involved in viral recognition. The response was independent of protein kinase R (Fig. 2b, left), suggesting that dsRNA was not passively leaking out of DC phagosomes into the cytosol²¹. MyD88, an adaptor downstream of many TLRs but not required for TLR3 signalling, was also not essential (Fig. 2b, right). In contrast, TLR3-deficient CD8⁺ DCs were completely unable to respond to poly I:C-loaded cells although they responded normally to the control stimulus CpG, which signals via TLR9 (Fig. 2c). Similarly, Tlr3^-/- CD8⁺ DCs were no longer activated by EMCV-infected or SFV-infected cells (Fig. 2c). We conclude that TLR3-dependent recognition of dsRNA in phagosomes can mediate activation of CD8⁺ DCs by poly I:C-loaded or virus-infected cells.

To assess the efficiency with which virally infected cells induce CTL cross-priming, we immunized mice with cells infected with 'suicide' SFV encoding a non-secreted form of ovalbumin (OVA) as a model antigen. Because the infected cells cannot release viral progeny (see above), there is no risk of infecting APCs in vivo and priming CTLs directly²². As a further precaution, we treated the infected cells with trypsin and succinate, which inactivate any infectious virus particles that remain bound to the cell surface (unpublished observations). As an alternative approach, we also used a virus-free system, immunizing mice with cells loaded with OVA protein in the presence or absence of poly I:C to mimic viral infection. Electroporation resulted in the loading of 50–250 ng OVA per 10⁶ cells, with over 80% of the cells expressing immunodetectable levels of protein (see Supplementary Fig. 5). Co-loading of poly I:C had no effect on antigen levels (see Supplementary Fig. 5a, b). SFV-OVA infection resulted in the expression of 100–200 ng OVA per 10⁶ cells (see Supplementary Fig. 5c), with 30–50% of cells showing positive staining for the protein (Supplementary Fig. 4b). Injecting 10⁶ OVA-electroporated or SFV-OVA-infected cells is therefore equivalent to administering less than 300 ng OVA protein. Remarkably, in cell-associated form with poly I:C, this low level of antigen was sufficient to induce robust cross-priming responses in naïve mice as determined by the expansion of OVA-specific endogenous CD8⁺ T cells or in vivo cytotoxicity (Fig. 3 and Supplementary Fig. 6a). Similarly, strong cross-priming of endogenous OVA-specific CTLs was seen after immunization with SFV-OVA-infected Vero cells (Fig. 3 and Supplementary Fig. 6a). In contrast, immunization with Vero cells loaded with OVA alone led to no detectable increase in CTL number and only low levels of OVA-specific in vivo cytotoxicity compared with control mice (Fig. 3 and Supplementary Fig. 6a). The adjuvanticity of poly I:C co-loading was also apparent in mice that had received a small number of naïve T cells from OT-I mice to increase the frequency of precursor CTLs before immunization (see Supplementary Fig. 6b). Indeed, the adjuvant effect of cell-associated poly I:C compared favourably with that of anti-CD40 monoclonal antibodies (see Supplementary Fig. 6b), widely regarded as the 'gold standard' for inducing CTL responses. We conclude that virally infected cells are potent inducers of cross-priming in vivo and that their immunogenicity can be mimicked by introducing dsRNA into uninfected cells.

Figure 3: dsRNA or viral infection promotes in vivo CTL cross-priming to cell-associated antigen.

Naive C57BL/6 mice were immunized with Vero cells loaded with OVA poly I:C or infected with SFV-OVA. Mice were injected with CFSE-labelled target cells on day 6 post-immunization. Splenocytes were isolated the following day and analysed for Thy1.2⁺ tetramer⁺ CD8⁺ cells (left panel) and the persistence of target cells (right panel). Data points represent individual mice.

High resolution image and legend (45K)

To relate the CTL cross-priming data to the TLR3-dependence of CD8⁺ DC activation, we tested for cross-priming in TLR3-deficient mice. Although these mice mount normal CTL responses to many viruses²³, they showed impaired responses to SFV-OVA-infected Vero cells as determined by CTL activity or IFN- ELISPOT assays (Fig. 4a, b). The impairment in cross-priming was absolute when OVA + poly I:C-loaded cells were used as the immunogen (Fig. 4b). To specifically map the defect in cross-priming to lack of TLR3 expression by APCs, we reconstituted lethally irradiated animals with bone marrow from Tlr3^-/- donors or Tlr3^+/+ littermate controls, transferred OT-I T cells and immunized with SFV-OVA-infected or OVA + poly I:C-loaded cells. The increased cell immunogenicity conferred by poly I:C loading was completely lost in the Tlr3^-/- chimaeras (Fig. 4c). Similarly, a reduced response was seen in Tlr3^-/- chimaeras immunized with SFV-OVA cells (Fig. 4d). As controls, Tlr3^-/- chimaeras mounted normal responses against OVA-loaded Vero cells co-injected with CpG + anti-CD40 (see Supplementary Fig. 7). Given that both the OT-I cells and the non-haematopoietic compartment in these mice are TLR3-sufficient, these data suggest that TLR3 expression by radiosensitive APCs is involved in in vivo cross-priming of CD8⁺ T cells against dsRNA-bearing, virally infected cells.

Figure 4: TLR3 dependence of CTL cross-priming against dsRNA-loaded or virally infected cells.

a, WT (Tlr3^+/+, upper panels) and Tlr3^-/- (lower panels) mice were immunized with SFV-OVA-infected Vero cells or PBS. CTL activity was measured by chromium release assay after re-stimulation ex vivo using target cells pulsed with OVA peptide (filled circles) or left unpulsed (open circles). b, WT (open circles) and Tlr3^-/- (filled circles) mice were immunized with Vero cells loaded with OVA poly I:C, infected with SFV-OVA or mock-treated. OVA-specific CTLs were measured by IFN- ELISPOT on day 9. c, d, Chimaeric mice reconstituted with bone marrow from Tlr3^+/+ (open circles) or Tlr3^-/- (filled circles) mice were injected with OT-I T cells and immunized the following day with Vero cells loaded with OVA poly I:C (c) or Vero cells infected with SFV-OVA (d). Numbers of OT-I T cells and in vivo specific killing of target cells were measured on day 14. Lines (a) and data points (b–d) represent individual mice.

High resolution image and legend (57K)

It is widely accepted that T cell priming requires DC activation by signals associated with the presence of infection. For viruses, direct infection can lead to DC activation via cytosolic pattern recognition of dsRNA intermediates of viral replication²¹. However, viruses that do not have tropism for DCs, such as lymphocytic choriomeningitis virus (LCMV) Armstrong²¹, EMCV or SFV (see Supplementary Fig. 4a, b), or viruses such as influenza that sequester dsRNA²¹, may fail to trigger this pathway. Such viruses can nevertheless be recognized through TLRs specific for viral genomes provided that enough viral particles gain access to the endosomal compartment of DCs, where those receptors are localized²⁴. However, many viruses are only ever 'seen' by DCs in cell-associated form. Here, we show that DCs use TLR3 to detect cell-associated viral dsRNA and that this receptor plays a role in vivo in cross-priming against virally infected cells. Our data provide a rationale for the existence of a pattern-recognition receptor for dsRNA that faces into the endosomal compartment, and also explain the apparent restriction of TLR3 expression to those subsets of APCs that avidly phagocytose dying cells^{15, 25}. Furthermore, our results offer a possible explanation for the controversial role of TLR3 in antiviral responses. Indeed, TLR3-deficient mice fail to show increased susceptibility to many viral infections²³ and display altered responses to others²⁶. On the basis of our results, one might predict that TLR3-dependence relates directly to the relative contribution of cross-priming versus direct priming to the antiviral response, which is a matter of considerable debate²⁷. Nevertheless, even in infections that strongly depend on cross-priming, the contribution of TLR3 is unlikely to be absolute. Indeed, Tlr3^-/- mice still mount residual CTL responses to SFV-OVA-infected cells (Fig. 4), clearly indicating the presence of alternative pathways for cross-priming during viral infection.

TLR3 signalling augments cross-presentation by DCs and leads to upregulation of co-stimulatory molecules and production of immunomodulatory cytokines²⁸. All these facets of DC activation are probably responsible for the increased cross-priming response to dsRNA-loaded cells in vivo. However, the upregulation of co-stimulatory molecules may be secondary to IFN-/ induction by TLR3 and autocrine signalling through the IFN-/ receptor²⁹. Notably, virally infected cells themselves also produce IFN-/ and activate DCs to promote CTL cross-priming¹³. We have used xenogeneic cells in our experiments in order to eliminate the effect of IFN-/ derived from the infected cells and to isolate the contribution of TLR3. This may represent the situation that arises during infection with viruses that block IFN-/ production³⁰. In other viral infections, TLR3 recognition of cell-associated dsRNA, together with IFN-/ produced by infected cells, are likely to act synergistically to promote DC activation. Thus, in an autologous setting, the cross-priming response to dsRNA-bearing cells is likely to be even more potent than that seen here with xenogeneic cells. Tumour cells loaded with dsRNA could therefore constitute potent vaccines for cancer immunotherapy.

References

1. Heath, W. R. et al. Cross-presentation, dendritic cell subsets, and the generation of immunity to cellular antigens. Immunol. Rev. 199, 9–26 (2004) | Article | PubMed | ISI | ChemPort |
2. Fonteneau, J. F. et al. Activation of influenza virus-specific CD4 + and CD8 + T cells: a new role for plasmacytoid dendritic cells in adaptive immunity. Blood 101, 3520–3526 (2003) | Article | PubMed | ISI | ChemPort |
3. Salio, M., Palmowski, M. J., Atzberger, A., Hermans, I. F. & Cerundolo, V. CpG-matured murine plasmacytoid dendritic cells are capable of in vivo priming of functional CD8 T cell responses to endogenous but not exogenous antigens. J. Exp. Med. 199, 567–579 (2004) | Article | PubMed | ISI | ChemPort |
4. Allan, R. S. et al. Epidermal viral immunity induced by CD8alpha + dendritic cells but not by Langerhans cells. Science 301, 1925–1928 (2003) | Article | PubMed | ISI | ChemPort |
5. Smith, C. M. et al. Cutting edge: conventional CD8alpha(+ ) dendritic cells are preferentially involved in CTL priming after footpad infection with herpes simplex virus-1. J. Immunol. 170, 4437–4440 (2003) | PubMed | ISI | ChemPort |
6. Belz, G. T. et al. Cutting edge: conventional CD8alpha(+ ) dendritic cells are generally involved in priming CTL immunity to viruses. J. Immunol. 172, 1996–2000 (2004) | PubMed | ISI | ChemPort |
7. den Haan, J. M., Lehar, S. M. & Bevan, M. J. CD8(+ ) but not CD8(-) dendritic cells cross-prime cytotoxic T cells in vivo. J. Exp. Med. 192, 1685–1696 (2000) | Article | PubMed | ISI | ChemPort |
8. Valdez, Y. et al. Major histocompatibility complex class II presentation of cell-associated antigen is mediated by CD8alpha + dendritic cells in vivo. J. Exp. Med. 195, 683–694 (2002) | Article | PubMed | ISI | ChemPort |
9. Iyoda, T. et al. The CD8(+ ) dendritic cell subset selectively endocytoses dying cells in culture and in vivo. J. Exp. Med. 195, 1289–1302 (2002) | Article | PubMed | ISI | ChemPort |
10. Schulz, O. & Reis e Sousa, C. Cross-presentation of cell-associated antigens by CD8⁺ dendritic cells is attributable to their ability to internalise dead cells. Immunology 107, 183–189 (2002) | Article | PubMed | ISI | ChemPort |
11. Belz, G. T. et al. The CD8alpha(+ ) dendritic cell is responsible for inducing peripheral self-tolerance to tissue-associated antigens. J. Exp. Med. 196, 1099–1104 (2002) | Article | PubMed | ISI | ChemPort |
12. Liu, K. et al. Immune tolerance after delivery of dying cells to dendritic cells in situ. J. Exp. Med. 196, 1091–1097 (2002) | Article | PubMed | ISI | ChemPort |
13. Le Bon, A. et al. Cross-priming of CD8 + T cells stimulated by virus-induced type I interferon. Nature Immunol. 4, 1009–1015 (2003) | Article |
14. Alexopoulou, L., Holt, A. C., Medzhitov, R. & Flavell, R. A. Recognition of double-stranded RNA and activation of NF-kappaB by Toll-like receptor 3. Nature 413, 732–738 (2001) | Article | PubMed | ISI | ChemPort |
15. Edwards, A. D. et al. Toll-like receptor expression in murine DC subsets: lack of TLR7 expression by CD8⁺ DC correlates with unresponsiveness to imidazoquinolines. Eur. J. Immunol. 33, 827–833 (2003) | Article | PubMed | ISI | ChemPort |
16. Kaufman, R. J. Double-stranded RNA-activated protein kinase mediates virus-induced apoptosis: a new role for an old actor. Proc. Natl Acad. Sci. USA 96, 11693–11695 (1999) | Article | PubMed | ChemPort |
17. Harshyne, L. A., Watkins, S. C., Gambotto, A. & Barratt-Boyes, S. M. Dendritic cells acquire antigens from live cells for cross-presentation to CTL. J. Immunol. 166, 3717–3723 (2001) | PubMed | ISI | ChemPort |
18. Helwig, F. & Schmidt, E. A filter-passing agent producing interstitial myocarditis in anthropoid apes and small animals. Science 102, 31–33 (1945) | ISI |
19. Strauss, J. H. & Strauss, E. G. The alphaviruses: gene expression, replication, and evolution. Microbiol. Rev. 58, 491–562 (1994) | PubMed | ISI | ChemPort |
20. Smerdou, C. & Liljestrom, P. Two-helper RNA system for production of recombinant Semliki forest virus particles. J. Virol. 73, 1092–1098 (1999) | PubMed | ISI | ChemPort |
21. Diebold, S. S. et al. Viral infection switches non-plasmacytoid dendritic cells into high interferon producers. Nature 424, 324–328 (2003) | Article | PubMed | ISI |
22. Freigang, S., Egger, D., Bienz, K., Hengartner, H. & Zinkernagel, R. M. Endogenous neosynthesis vs. cross-presentation of viral antigens for cytotoxic T cell priming. Proc. Natl Acad. Sci. USA 100, 13477–13482 (2003) | Article | PubMed | ChemPort |
23. Edelmann, K. H. et al. Does Toll-like receptor 3 play a biological role in virus infections? Virology 322, 231–238 (2004) | Article | PubMed | ISI | ChemPort |
24. Crozat, K. & Beutler, B. TLR7: A new sensor of viral infection. Proc. Natl Acad. Sci. USA 101, 6835–6836 (2004) | Article | PubMed | ChemPort |
25. Muzio, M. et al. Differential expression and regulation of toll-like receptors (TLR) in human leukocytes: selective expression of TLR3 in dendritic cells. J. Immunol. 164, 5998–6004 (2000) | PubMed | ISI | ChemPort |
26. Tabeta, K. et al. Toll-like receptors 9 and 3 as essential components of innate immune defense against mouse cytomegalovirus infection. Proc. Natl Acad. Sci. USA 101, 3516–3521 (2004) | Article | PubMed | ChemPort |
27. Melief, C. J. Regulation of cytotoxic T lymphocyte responses by dendritic cells: peaceful coexistence of cross-priming and direct priming? Eur. J. Immunol. 33, 2645–2654 (2003) | Article | PubMed | ISI | ChemPort |
28. Reis e Sousa, C. Toll-like receptors and dendritic cells: for whom the bug tolls. Semin. Immunol. 16, 27–34 (2004) | Article | PubMed | ChemPort |
29. Hoebe, K. & Beutler, B. LPS, dsRNA and the interferon bridge to adaptive immune responses: Trif, Tram, and other TIR adaptor proteins. J. Endotoxin Res. 10, 130–136 (2004) | Article | PubMed | ISI | ChemPort |
30. Katze, M. G., He, Y. & Gale, M. Viruses and interferon: A fight for supremacy. Nature Rev. Immunol. 2, 675–687 (2002) | Article |

Supplementary Information

Supplementary information accompanies this paper.

Acknowledgements

This work was funded by Cancer Research UK (C.R.S.) and by the Swedish Research Council and the EU program (P.L.). We thank I. Kerr for providing EMCV and anti-EMCV antiserum, L. van Dinten for suggestions on 'suicide' virus models and L. Kostic for technical assistance. We are grateful to R. Germain, I. Kerr and members of the Immunobiology Laboratory, Cancer Research UK, for advice and critical review of the manuscript. R.A.F. is an investigator of the Howard Hughes Medical Institute, M.A.N. is supported by an EMBO long-term fellowship and Y.T.A. is supported by the Nakatomi Foundation.

Competing interests statement

The authors declare no competing financial interests.

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.