INSERM U345, Institut Necker, 75015 Paris, France. rocha@necker.fr
Maintenance of a memory cell population is thought to be independent of antigen. However, it has now been shown that unless memory cells get some signals via MHC, they become functionally defective.
Induction of efficient long-term immune memory is the aim for all vaccination protocols. Whereas some vaccines induce considerable protection and have been used to control or eradicate major pathogens, the development of efficient vaccines against many current human pandemics remains elusive. Identification of the requirements for memory maintenance is, therefore, a fundamental issue, not only for our understanding of immune responses in general but also for creating effective strategies for vaccine design. The article by Kassiostis et al.1 in this issue of Nature Immunology sheds new light on this issue.
In mice, memory T cell survival does not require the persistence of cognate antigen2,
3,
4. Some studies have indicated that T cell receptor (TCR) triggering4,
5,
6 by major histocompatibility complex (MHC) proteins4,
5 is required, but others have demonstrated the survival of memory T cells even after transfer into MHC-deficient mice7,
8. These latter experiments led to the concept that—in contrast to naïve cells4—memory cells, once generated, become independent of further TCR-mediated signals7,
8. This concept was received by some, especially clinical researchers, with skepticism9. Indeed, serial analyses of antigen-specific T cells in patients infected by viruses (HIV in particular) show a correlation between viral load and the percentage of antigen-specific cells in the blood10. HIV-specific memory cell counts fall sharply when anti-retroviral therapy is initiated, matching the fall in viral load; in some circumstances, specific memory T cells are undetectable in patients with undetectable viral load10. These results suggest that the size of the memory T cell pool in humans is highly dependent on the persistence of cognate antigen.
How can such conflicting findings be reconciled? Kassiostis et al.1 used an experimental system in which they were able to generate memory CD4+ T cells and maintain them in different environments. Two different monoclonal TCR-transgenic memory CD4+ T cell lines were generated and maintained either in the absence of MHC class II antigens (conditions in which TCR triggering should not take place) or in the presence of allogeneic MHC class II antigens. In this allogeneic environment, the cognate antigen cannot be recognized, although some form of MHC-mediated TCR triggering may still occur. Kassiostis et al. compared the number and functional properties of such memory cells1 (Fig. 1). They found that the absence of MHC contact did not modify the survival of these cells in vivo or their capacity to proliferate or express intracytoplasmic cytokines after in vitro stimulation by peptide-loaded dendritic cells1. In other words, when tested with the methods most frequently used to study memory T cell function, these cells performed well.
Figure 1. Survival signals to maintain memory function.
Memory T (TM) cells maintained in the presence of allo-MHC or no MHC were tested (a) in vitro and (b) in vivo.
These results contrast starkly with the memory behavior observed in vivo (Fig. 1). In particular, memory CD4+ T cells maintained in an environment lacking MHC class II antigens were unable to collaborate with B cells to induce immunoglobulin secretion. They were also unable to reject skin transplants expressing the antigen. Rejection requires processing and presentation of skin antigens by the graft Langerhans cells. The failure of memory T cells maintained in the absence of MHC class II to induce skin rejection indicates that they are unable to recognize and respond to antigens processed by dendritic cells in vivo.
What, then, are the main messages of this article1? The first is that the in vitro tests most frequently used to examine memory T cell behavior are not sufficiently sensitive. They may fail to reveal major differences in the functional capacities of memory T cells. This idea is in line with published data11,
12, which show that cytokine expression after in vitro stimulation does not reflect in vivo cytokine expression. The data also extend the scope of functional deficiencies that may be ignored. Because most of the in vitro assays usually used are so crude, T cell functional behavior during immune responses may be far more heterogeneous than currently appears. Functional differences among primed cells may explain why some infections are controlled, whereas others are not. New and more accurate tests are therefore required for the study of T cell function if we are to understand the heterogeneity of immune responses.
Second, Kassiostis et al.1 show that although memory CD4+ T cells can survive in the absence of stimulation, they lose the functional properties of memory cells. Some tests show that they are even less efficient than naïve T cells. For example, although naïve T cells can reject skin graphs (albeit slowly), these "deprived" memory cells cannot. This indicates that functional memory (the only parameter that really matters) needs some kind of periodic triggering by MHC. They also show that the required conditions for maintaining optimal memory function have yet to be determined. More complex approaches to the analysis of T cell function are clearly required to achieve this. What causes optimal triggering? Is syngeneic MHC better than allogeneic MHC? And might the cognate antigen be even more efficient in this respect (as defended by some)9? We should not rush to embrace this latter hypothesis. Efficient secondary immune responses do not require high frequencies of memory T cells. Memory T cells maintained in the absence of cognate antigen are so efficient that a small number can mount potent secondary responses in vivo11. The persistence of cognate antigen, on the other hand, may lead to anergy or exhaustion13. The "window" in which optimal memory function is maintained and tolerance is avoided may be a very narrow one.
