News & Views | Published:

Cancer immunology

Cat and mouse games

The immune system is intimately involved in how tumours develop. But how do tumours avoid being killed by immune responses? It seems that in some instances they can lull immune cells into a false sense of security.

A troubled relationship exists between the immune system and tumours — not unlike that between the cartoon characters Tom and Jerry. Tom's disposition is to catch and devour mice, but he hardly ever succeeds in capturing Jerry, who has just too many tricks up his sleeve. On page 141 of this issue Willimsky and Blankenstein1 report a novel ploy used by tumours to evade the immune system. It seems that sporadic tumours in mice, against which the immune system initially reacts, nevertheless manage to spread by moulding the immune cells that are most effective against them — the killer T cells — into a state that tolerates them.

The report adds fuel to a controversy that has been raging for some time, namely whether the development of sporadic, non-viral tumours is really affected by so-called immunoediting. The theory of immuno-editing posits that tumours eliciting strong immune responses from the T cells of their host will experience a darwinian selection pressure to avoid destruction by the immune system. So, they evolve ‘immune escape’ or ‘sculpted’ variants, usually by losing or downregulating the molecules that alerted the immune system in the first place2,3.

In Willimsky and Blankenstein's experiment, however, it is not the tumour cells that adapt to immune attack — rather, the immune system is coaxed into inertia. The authors genetically engineered an ingenious mouse model, which contained a cancer-promoting gene of viral origin (encoding the ‘SV40 large T’ protein). The gene was controlled so that it was activated only rarely and in random tissues. The mice thus developed rare sporadic tumours, all expressing SV40 large T. Although this protein initially provoked an immune response in the mice, in the course of tumour development the mice became immunologically tolerant of it. However, the tumours were vigorously rejected when transplanted into identical mice without tumours — so they could still elicit a tumour-killing immune response1 — and they showed no obvious signs of immunoediting1,2.

Curiously, both sides of the controversy concur that both virus-induced and sporadic tumours do elicit immune responses, and that T-cell responses protect against viral carcinogenesis. An example of the latter is the markedly increased incidence of cancers induced by Epstein–Barr virus in AIDS patients and in immunosuppressed recipients of organ transplants4. These tumours also express a much wider array of viral antigens4. But what distinguishes the immune response against virus-induced cancers from that against sporadic tumours? After all, Willimsky and Blankenstein's model was based on sporadic carcinogenesis, but using the viral SV40 large T protein. This protein is one of the main targets of successful T-cell responses against tumours induced by intact SV40, the virus from which the large T protein is derived. Indeed, the efficient T-cell reactions against this antigen are instrumental in the prevention of tumour development in mice that have been injected with intact SV40 (ref. 5).

Willimsky and Blankenstein offer little explanation for the striking difference in the protective power of immune responses against the identical viral antigen. And they conclude from their results that cancers that are not induced by viruses do not undergo immuno-editing. But they fail to take into account the extensive documentation from others that such tumours are in fact subject to selective pressure from immune responses2,3.

How can these seemingly contradictory facts and widely divergent views be reconciled? Two points are worth emphasizing here. First, tumours need to be sensed by an exquisitely sensitive sentinel system, the dendritic cell network (Fig. 1). These cells not only sense the tumours, they also direct the subsequent immune response. Dendritic cells pick up material from dead or dying tumour cells, and ‘present’ it to T cells. For proper arousal of killer T cells, the dendritic cells must be activated by secondary ‘danger’ signals, some of which can be emitted by tumours. The presentation of the tumour debris by the activated dendritic cell causes T cells to multiply and attack the remaining tumour cells6. However, if the dendritic cells are not activated, the T cells will take little notice and can become tolerant of the tumour6,7,8.

Figure 1: Escaping the immune system — a model.

After initial growth, tumours usually shed some immunogenic material from dead or dying tumour cells. This debris is picked up by dendritic cells, which transport it to the lymph node and ‘present’ it to T cells. The subsequent immunological events are determined by the manner in which the tumour is perceived by the dendritic cell network. a, If the tumour, apart from shedding debris, also emits ‘danger’ signals such as stress proteins, the dendritic cells will be activated. These activated cells present the tumour debris to the T cells, eliciting a robust response and causing the T cells to multiply and kill tumour cells. The only way for tumour cells to survive is to escape by immunoediting2,7. b, If the tumour manages to masquerade as healthy tissue, giving off no danger signals, the dendritic cells are not activated. The T cells therefore tolerate the tumour material presented to them, and do not become killers9. Tumours capable of such tolerance induction do not need immunoediting to escape from immune attack. Tumours that are induced by viral infection are more likely to fall into the first category; Willimsky and Blankenstein's1 mouse model seems to produce tumours that fall into the second.

Second, the natural ability of tumours to activate dendritic cells is likely to vary widely, from no stimulatory activity to considerable stimulatory action through endogenous danger signals such as overexpressed stress proteins and interferons (proteins that were named for their capacity to interfere with viral infections but are also produced in response to non-viral tumours)6,9. So an important distinction should be made between tumours whose growth is associated with sufficient dendritic-cell activation to generate killer T cells targeted against them, and tumours that do not (Fig. 1). Immunoediting should occur only in the first category, because escape is needed only in the face of immune attack.

Although virus-associated tumours are more likely to cause dendritic-cell activation than sporadic tumours, and are therefore more likely to be subjected to immunoediting, this model seems to be too simple. Many sporadic, non-viral tumours can apparently arouse enough of the body's danger signals to become subject to immune attack and undergo immunoediting2,3,9. Conversely, most tumour viruses fail to incite sufficient danger signals in susceptible individuals and consequently can establish the persistent viral infection associated with a high risk of cancer10.

So, the bad news is that cancer cells can avoid robust T-cell-mediated immune attack, either by failing to arouse the immune system, thereby causing tolerance1, or by immuno-editing2,9. The good news is that T cells can be turned into robust killers by the vigorous activation of dendritic cells using a variety of molecules8, and this may be of use in cancer vaccines6,8 and in T-cell immunotherapy11.


  1. 1

    Willimsky, G. & Blankenstein, T. Nature 437, 141–146 (2005).

  2. 2

    Dunn, G. P. et al. Nature Immunol. 3, 991–998 (2002).

  3. 3

    Khong, H. T. & Restifo, N. P. Nature Immunol. 3, 999–1005 (2002).

  4. 4

    Straathof, K. C. M. et al. Oncologist 8, 83–98 (2003).

  5. 5

    Newmaster, R. S., Mylin, L. M., Fu, T. -M. & Tevethia, S. S. Virology 244, 427–441 (1998).

  6. 6

    Melief, C. J. M. Nature Immunol. 6, 543–544 (2005).

  7. 7

    Probst, H. C. et al. Nature Immunol. 6, 280–286 (2005).

  8. 8

    Dumortier, H. et al. J. Immunol. 175, 855–863 (2005).

  9. 9

    Dunn, G. P. et al. Nature Immunol. 6, 722–729 (2005).

  10. 10

    De Jong, A. et al. Cancer Res. 64, 5449–5455 (2004).

  11. 11

    Klebanoff, C. A. et al. Proc. Natl Acad. Sci. USA 102, 9571–9576 (2005).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.