News & Views | Published:


It takes more than two to tango

Naturevolume 409pages3132 (2001) | Download Citation


In vivo studies of a pair of co-stimulatory molecules in the immune system of mice may further our understanding of allergic reactions and inflammatory immune responses in humans.

Most T cells of the immune system recognize antigens — substances that stimulate an immune response — as protein fragments. These fragments must be presented to T cells by molecules on the surface of another type of cell, an antigen- presenting cell (APC). But this recognition event alone is not enough to fully activate the T cell. Simultaneous 'co-stimulatory' signals from the APC are needed before the T cell can proliferate and become specialized to perform a particular function, such as secreting intercellular signalling molecules called cytokines. On pages 97–109 of this issue1,2,3, three groups investigate a receptor protein called ICOS, which occurs on T cells and recognizes a co-stimulatory signal on APCs. All three groups agree that experimentally inactivating the ICOS gene in mice profoundly impairs the animals' ability to produce certain types of cytokines and antibody molecules.

The first co-stimulatory pair of molecules to be identified were the receptor protein CD28 — expressed continuously on all mouse and many human T cells — and its two binding partners, CD80 and CD86. The expression of these binding partners, or ligands, is not continuous but rather can be induced on all APCs4 (Fig. 1). Together with signalling through the T cell's antigen receptors, this co-stimulatory set enhances the production of the cytokine interleukin-2, which encourages the growth and differentiation of T cells. Subsequent work uncovered a second molecule on the T-cell surface that recognizes CD80 and CD86. The expression of this molecule, called CTLA-4, is inducible and imposes negative feedback on the process of T-cell activation.

Figure 1: Co-stimulatory signals and immune responses.
Figure 1

a, A dendritic cell (a type of antigen-presenting cell) is activated to express the co-stimulatory molecules CD80 and CD86. It can then present antigen to a T cell, which is activated by signalling through its antigen receptor and through CD28, the receptor for CD80/86. b, The activated T cell expresses all of its inducible co-stimulatory-signal receptors, such as ICOS. c, When the T cell next encounters a B cell that has become an antigen-presenting cell, the interactions of the co-stimulatory sets of molecules augment the production of cytokines (interleukins 4 and 13) by the T cell and of antibodies by the B cell (d). The interaction between ICOS and B7RP-1 is particularly potent in stimulating immunoglobulin E production, important in allergic reactions. Blocking this might be useful in treating asthma. Three new papers1,2,3 show that inactivating the ICOS gene in mice inhibits production of these cytokines and antibody, among other effects. Molecules in red are expressed only after cell activation.

A completely different molecular pair was then also found to be essential for co-stimulation5. In this case, the co-stimulatory signal (CD40) is continuously expressed on the APC, and the expression of its receptor (CD154) can be induced on the T cell (Fig. 1). Signalling through this co-stimulatory pair creates a strong positive-feedback loop, augmenting the expression of CD80 and CD86 on the APC. Many other co-stimulatory molecules have since been described. One of the most recently discovered pairs — consisting of ICOS and its binding partner, B7RP-1 — is another positive regulatory set.

Why is co-stimulation so complex? Presumably, each molecular pair has some important, unique function and so needs to be regulated independently of the other sets. A major clue to the function of ICOS and its partner lies in where and how these molecules are expressed. ICOS is expressed on the T cell in response to signalling through CD28 and the T-cell antigen receptor6. More important, its binding partner has a unique expression pattern. B7RP-1 is found only on the types of APCs called B cells and macrophages, and not on the major category of APC, a particular subset of dendritic cells7. Although other subsets of dendritic cells still need to be examined, these observations suggest that the ICOS–B7RP-1 pair is not involved in the initial activation of the T cell, but rather is important later on, when the B cells and macrophages are activated (Fig. 1).

The effects of inactivating ICOS in mice, described by Dong et al.1, McAdam et al.2 and Tafuri et al.3, are consistent with this idea. The 'knockout' mice have few germinal centres — the lymphoid factories where B cells manufacture antibody molecules. And certain types of antibodies, such as immunoglobulin E, are not generated at all in these mice. This can be partly explained by the decrease in production of specific cytokines (such as interleukins 4 and 13) by T cells. But the fact that there is decreased production of almost all antibody types suggests a more general defect.

One potential explanation is a slight decrease in the production of interleukin-2, described by Dong et al.1 in in vitro analyses. But it might be more fruitful in the future to study the expression of chemokine receptors and their soluble ligands, chemokines. The chemokine-receptor family of cell-surface molecules, involved in the migration of immune cells during an immune response, is essential for the formation of germinal centres8. ICOS is expressed mainly on activated T cells that come into germinal centres, and its stimulation might amplify chemokine production to recruit more activated B and T cells to this site. In support of this idea, a 'transgenic' mouse expressing large amounts of soluble B7RP-1 has opposite characteristics to the knockout mouse — enlarged lymphoid tissue, with many germinal centres and antibody-forming cells7.

Alternatively, perhaps ICOS and its partner are mainly needed to augment antibody production by B cells: McAdam et al.2 found that the knockout defects can be partially overcome by strongly activating the other co-stimulatory pathways. And soluble ICOS has no effects on the antiviral responses of cytotoxic T cells9. Effects on the activation of macrophages remain to be tested.

As well as their fundamental importance, these results1,2,3 have implications for the study of human diseases. The production of the immunoglobulin E antibody is an early feature of allergic responses to antigens such as pollen or dust mites. So the diminished production of this antibody in mice lacking ICOS would be expected to alleviate allergic conditions. Indeed, one recent study showed that administering a soluble form of ICOS — which acts as a decoy, preventing B7RP-1 from binding to ICOS on T cells — reduced the lung obstruction in an animal model of asthma10. This effect is best explained by another interesting feature of the expression pattern of B7RP-1. Inflammatory cytokines such as tumour-necrosis factor-α induce the expression of B7RP-1 on fibroblast cells, found in the connective tissue of the lungs11. Binding of this B7RP-1 to ICOS on nearby T cells might encourage these cells — when they recognize their antigen in the environment of the lung — to make cytokines such as interleukins 4 and 13. This would in turn encourage B cells to secrete immunoglobulin E. The soluble ICOS molecule would block this process.

On the opposite pole, cytokines such as interleukins 4 and 13 can also reduce inflammatory immune responses mediated by T cells. Indeed, when Dong et al.1 inactivated the ICOS gene in mice immunized to show some of the symptoms of human multiple sclerosis, they found that this inflammatory disorder worsened. All in all, it seems that this co-stimulatory pair is well worth further investigation in connection with several human immune-mediated diseases.


  1. 1

    Dong, C. et al. Nature 409, 97–101 (2001).

  2. 2

    McAdam, A. J. et al. Nature 409, 102–105 (2001).

  3. 3

    Tafuri, A. et al. Nature 409, 105–109 (2001).

  4. 4

    Abbas, A. K. & Sharpe, A. H. Nature Med. 5, 1345–1346 (1999).

  5. 5

    Grewal, I. S. & Flavell, R. A Annu. Rev. Immunol. 16, 111–135 (1998).

  6. 6

    Hutloff, A. et al. Nature 397, 263–266 (1999).

  7. 7

    Yoshinaga, S. K. et al. Nature 402, 827–832 (1999).

  8. 8

    Cyster, J. Science 286, 2098–2102 ( 1999).

  9. 9

    Kopf, M. et al. J. Exp. Med. 192, 53–61 (2000).

  10. 10

    Coyle, A. J. et al. Immunity 13, 95–105 (2000).

  11. 11

    Swallow, M. M., Wallin, J. J. & Sha, W. C. Immunity 11, 423– 432 (1999).

Download references

Author information


  1. Laboratory of Cellular and Molecular Immunology, National Institutes of Health, National Institute of Allergy & Infectious Diseases, 9000 Rockville Pike, Bethesda, 20892-0420, Maryland, USA

    • Ronald H. Schwartz


  1. Search for Ronald H. Schwartz in:

Corresponding author

Correspondence to Ronald H. Schwartz.

About this article

Publication history

Issue Date


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.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing