CD8+ T cells are a critical component of the adaptive immune response against infections and tumors. A current paradigm in immunology is that naive CD8+ T cells require CD28 costimulation, whereas memory CD8+ T cells do not. We show here, however, that during viral infections of mice, costimulation is required in vivo for the reactivation of memory CD8+ T cells. In the absence of CD28 costimulation, secondary CD8+ T cell responses are greatly reduced and this impairs viral clearance. The failure of CD8+ T cells to expand in the absence of CD28 costimulation is CD4+ T cell help independent and is accompanied by a failure to down-regulate Bcl-2 and by cell cycle arrest. This requirement for CD28 costimulation was shown in both influenza A and HSV infections. Thus, contrary to current dogma, memory CD8+ T cells require CD28 costimulation to generate maximal secondary responses against pathogens. Importantly, this CD28 requirement was shown in the context of real infections were multiple other cytokines and costimulators may be up-regulated. Our findings have important implications for pathogens, such as HIV and measles virus, and tumors that evade the immune response by failing to provide CD28 costimulation. These findings also raise questions about the efficacy of CD8+ T cell-based vaccines against such pathogens and tumors.

Although much is known about the initiation of immune responses, much less is known about what controls the effector phase. CD8+ T cell responses are believed to be programmed in lymph nodes during priming without any further contribution by dendritic cells (DCs) and Ag. In this study, we report the requirement for DCs, Ag, and CD28 costimulation during the effector phase of the CD8+ T cell response. Depleting DCs or blocking CD28 after day 6 of primary influenza A virus infection decreases the virus-specific CD8+ T cell response by inducing apoptosis, and this results in decreased viral clearance. Furthermore, effector CD8+ T cells adoptively transferred during the effector phase fail to expand without DC, CD28 costimulation, and cognate Ag. The absence of costimulation also leads to reduced survival of virus-specific effector cells as they undergo apoptosis mediated by the proapoptotic molecule Bim. Finally, IL-2 treatment restored the effector response in the absence of CD28 costimulation. Thus, in contrast to naive CD8+ T cells, which undergo an initial Ag-independent proliferation, effector CD8+ T cells expanding in the lungs during the effector phase require Ag, CD28 costimulation, and DCs for survival and expansion. These requirements would greatly impair effector responses against viruses and tumors that are known to inhibit DC maturation and in chronic infections and aging where CD28−/− CD8+ T cells accumulate.

CD28 costimulation is crucial for priming of CD8+ T cell responses to pathogens, however its role in the later phase of the primary and in the secondary immune responses is not clear. To dissect the role of CD28, C57Bl/6 mice were primed intranasally with influenza virus and then treated with a non-depleting blocking anti CD28 antibody at days 6 and 8 post infection. The frequency and absolute number of virus-specific CD8+ T cells were reduced in anti CD28 treated (3.9±0.4%; 1.9±0.5x105) vs. untreated mice (11.5±1.2%; 7.4±1.7 x105). We found that anti-CD28 treatment induces apoptosis of virus-specific CD8+ cells while it did not affect the number of actively dividing cells. To study the role of CD28 during secondary CD8+ T cell responses, C57Bl/6 mice were primed with flu virus strain PR8 and day 60 memory cells were transferred into C57Bl/6 or CD80−/−CD86−/− deficient mice. Mice were then rechallenged with flu virus strain X31 and 7 days post rechallenge CD80−/−CD86−/− deficient mice exhibited a reduction in virus-specific CD8+ T cells in the lung, compared to C57Bl/6 mice (1.2±0.7x106 vs. 5.8±1.1x106, respectively). Failure of memory CD8+ T cells to expand in the absence of CD28 costimulation is due to a failure to downregulate Bcl-2 and to cell cycle arrest. Thus, effector and memory CD8+ T cells require CD28 costimulation to generate optimal immune responses against pathogens.