The economy of T-cell memory: CD4⁺ recession in times of CD8⁺ stability?

Although systemic viral infection gives rise to robust responses from both CD8⁺ and CD4⁺ T cells, there are striking differences as well as similarities in the way the two classes of lymphocyte cells react. Insights into the patterns of response by each type of T cell could lead to the development of new therapeutics that enhance or suppress factors governing CD8⁺ and CD4⁺ T-cell immunity. (pages 913–919)

The CD8⁺ T-cell response to acute viral infection has three distinct phases: expansion, contraction and memory (Fig. 1). Over the past few years, new methods have been developed to quantify T-cell responses during each of these phases. These new assays have revealed powerful CD8⁺ T-cell responses against a variety of viral infections including lymphocytic choriomeningitis virus, vaccinia virus, vesicular stomatitis virus, murine γ-herpesvirus and polyoma virus in mice and Epstein–Barr virus, HIV and cytomegalovirus in humans. In this issue, Homann et al.¹ use new, sensitive intracellular cytokine-staining assays and major histocompatibility class II (MHC-II)-tetramer analysis to more thoroughly examine CD8⁺ and CD4⁺ memory following lymphocytic choriomeningitis infection in mice. Several similarities and differences were found between the two classes of lymphocyte cells during different phases of the T-cell response.

Figure 1: Differences in magnitude and longevity of CD4⁺ and CD8⁺ memory.

Stephen Horwitz

The CD8⁺ T-cell response to acute viral infection has three distinct phases (top). Upon initial encounter with the virus, small numbers of CD8⁺ T cells proliferate (∼15 cell divisions) and differentiate into cells that are cytotoxic or that produce cytokines to eliminate infected cells. Once the infection has been cleared, this expanded population of cytotoxic effector cells is reduced in number by controlled apoptosis during a 'contraction' phase that lasts for a few weeks. However, a subset of memory cells survives this purging process and is maintained at constant numbers over time. Previous studies have found that the size of this surviving subset is proportional to the original number of cells that had proliferated during the expansion phase (burst size). The CD4⁺ response (bottom) is smaller in magnitude and has a delayed contraction phase compared with the CD8⁺ response. Furthermore, whereas memory CD8⁺ levels are unwavering over the lifetime of mice, memory CD4⁺ levels decay with a half-life of ∼400 days. It remains to be determined what factors regulate memory CD4⁺ survival during these later phases.

Although much is known about the dynamics of CD8⁺ T-cell responses², the magnitude, specificity and duration of CD4⁺ T-cell responses have only recently been examined^3,4,5. Virus-specific CD4⁺ T cells help control infection by secreting interferons to inhibit viral replication, activating macrophages or driving B cells to produce antiviral antibody. Recent studies have shown that CD4⁺ T cells can enhance cytotoxic T-lymphocyte development through either activation of antigen-presenting cells or direct secretion of interleukin-2. In some cases, induction of CD8⁺ responses depends on CD4⁺ T cells, and in other cases—particularly chronic viral infections—maintenance of antiviral CD8⁺ responses appears to also depend on CD4⁺ T cells. Therefore, it is important to understand the relation between CD4⁺ and CD8⁺ responses.

Homann et al. found that although CD4⁺ responses were readily detected during the primary expansion phase, the magnitude of the CD8⁺ response vastly exceeded that of the CD4⁺ response (∼20-fold). This confirms previous studies in mice⁶ and in humans where, for example, more CD8⁺ than CD4⁺ clones appear after infection by Epstein–Barr virus. Both T-cell expansions were restricted, with virtually all of the CD8⁺ and CD4⁺ T-cell proliferation accounted for by virus-specific cells, thus excluding significant bystander proliferation or recruitment of cells. Although this study did not investigate how many potentially virus-reactive CD8⁺ or CD4⁺ T-cell clones exist in naive mice, it is clear that there are qualitative and quantitative differences between CD8⁺ and CD4⁺ T cells. For example, CD8⁺ T cells divide to a greater extent (15 divisions) and faster (6 h per division) than CD4⁺ T cells (9 divisions, 9–20 h per division)^1,2. An emerging pattern is that CD4⁺ responses require the engagement of costimulatory receptors for optimal activation and proliferation whereas CD8⁺ responses are less dependent upon these interactions⁶; so although these differences are likely due to the greater amount of antigen that CD8⁺ T cells encounter, they might also reflect the additional restraints on CD4⁺ proliferation such as costimulatory molecule expression by antigen-presenting cells. Together, these findings make it apparent that induction of CD4⁺ and CD8⁺ T-cell responses differ in both activation requirements and in magnitude.

During the contraction phase, around 90–95% of the virus-specific CD8⁺ T cells die from apoptosis during a 1–2-week period. This efficient purging of CD8⁺ T cells likely reflects the need to remove potentially dangerous cells that might cause immunopathology. Consistent with an earlier report⁵, Homann et al. found that approximately 50 days were required to remove a similar percentage of CD4⁺ T cells. It is unclear why the CD4⁺ contraction phase is more protracted. Recent evidence indicates that activated CD8⁺ T cells—but not CD4⁺ T cells—are selectively trapped in the liver and then undergo apoptosis⁷. Rather than trafficking to liver or intestinal 'T-cell graveyards', perhaps CD4⁺ T cells preferentially localize to sites that protect them from death, such as germinal centers. The survival of CD4⁺ T cells during this period might therefore reflect the longevity of germinal centers.

One of the interesting findings in this report is the striking difference between the memory phase of CD8⁺ and CD4⁺ T cells in immune mice: CD8⁺ memory was maintained at constant levels from days 25–900, whereas CD4⁺ memory waned over time with a half-life of approximately 400 days. The combined use of MHC-II–tetramer staining and staining for intracellular interferon-γ indicated that virus-specific CD4⁺ cells were truly reduced in number and had not simply changed cytokine profile (that is, Th1→Th2) or become non-responsive to antigen stimulation. Also, memory CD8⁺ T cells express higher levels of Bcl-2 than naive or activated cells⁸, and Homann et al. found lower Bcl-2 levels in CD4⁺ memory cells than in CD8⁺ memory cells. This observation suggests one possibility that the regulatory mechanisms modulating levels of memory are functioning at least in part through anti-apoptotic gene expression.

Interesting questions arise from this work. For example, why is the first contraction phase delayed for CD4⁺ T cells? Although CD8⁺ responses clearly show three phases with memory set at around 5–10% of the original burst size, it is unclear for CD4⁺ responses when the contraction phase ends and the memory phase begins. Levels of CD4⁺ memory tended to settle at 5% for a while but then continued to decrease to about 1% in aged mice. The decrease in CD8⁺ cell number during the contraction phase in this system reflects increased cell death, but the rules could be different for CD4⁺ T cells. Homann et al. examined memory in lymphoid organs, but could it be that there is CD4⁺ memory in other non-lymphoid tissues over time? It remains to be seen whether CD4⁺ T cells are dispersed from the spleen to other sites so that there is a repository for them elsewhere as reported recently in another experimental system⁹.

The contraction phase has an important role in setting levels of CD8⁺ memory. When B cells are missing or when CD40–CD40L interaction is blocked, the contraction phase is exaggerated and results in fewer memory cells⁶. However, once established, the longevity of CD8⁺ memory is unaffected by the absence of these interactions. Given that initial CD4⁺ T-cell responses are more dependent on costimulatory interactions than CD8⁺ T cells, it is conceivable that the establishment of CD4⁺ memory during the contraction phase is also more affected by these interactions than CD8⁺ memory. B cells in particular might have a key role in this process since CD4⁺ T cells and B cells are closely localized in lymphoid organs and cross-communicate through costimulatory molecules as well as cytokines. Evidence for this is provided by a recent study that demonstrated lower CD4⁺ memory in B-cell–deficient mice¹⁰. Proof that CD4⁺ memory is more sensitive to B-cell–dependent survival signals than CD8⁺ memory will come from studies in which both subsets are monitored in the same mice. Future studies focused on the contraction phase of the CD4⁺ response in the presence or absence of B cells should shed some light on what B-cell–dependent signals are involved in establishing and maintaining CD4⁺ memory.

It will be important to determine if loss of CD4⁺ memory is a general characteristic of aging on homeostasis of memory cells. For example, is there an overall reduction in the numbers of CD4⁺CD44^Hi memory cells during the aging process or is this reduction specific for the virus-specific memory cells studied here? Given that the analysis of Homann et al. suggests that levels of CD4⁺ memory are being surveyed and regulated long after immunization, what signals operate as much as two years after infection to control memory CD4⁺ T-cell survival or death? Do B-cell–dependent signals or costimulatory interactions continue to function during these very late periods? In some systems antigen is maintained over time in the form of antigen–antibody complexes on follicular dendritic cells. These complexes are thought to be picked up by B cells, processed and presented with MHC-II to memory CD4⁺ T cells. Could it be that these antigen depots are necessary to maintain CD4⁺ memory and that the gradual loss of memory seen by Homann's group is due to the gradual loss of these complexes over time? It has also been proposed that maintenance of memory T-cell numbers requires low-level homeostatic proliferation¹¹. Perhaps CD4⁺ and CD8⁺ memory cells sense different cytokines that affect their homeostatic proliferation. Do virus-specific CD4⁺ memory cells undergo less homeostatic proliferation than CD8⁺ memory cells? Alternatively, is the long-term survival of memory cells dictated by some 'differentiate-into-memory-cell' signal received during the initial expansion phase? Evidence for this is provided by a study showing that brief encounters with antigen are sufficient to differentiate CD8⁺ memory cells¹². Perhaps CD8⁺ T cells receive more of this signal than CD4⁺ T cells and are therefore better able to persist over long periods of time. Many of the interesting questions raised by this study can now be addressed with the advent of the sensitive assays used here.