Vaccine leads to memory loss

A promising experimental vaccine strategy, just entering clinical trials, displays a deleterious effect in mice. Use of antibodies to CD40 seems to clamp down on the long-term ability of T cells to respond to antigen (pages 354–360).

Can a vaccine be 'too efficient'? Can it provide such a strong stimulus to the immune system that the response is ultimately abortive rather than productive? In this issue, Berner et al.¹ use mouse models to show that a widely heralded approach to vaccine adjuvants, targeting of the CD40 receptor, can have an unexpected deleterious side effect—it abolishes long-term T-cell responsiveness toward tumor antigens. These findings dovetail with recent work by Bartholdy et al.², who came to similar conclusions in a viral infection model.

Together, these data show that our understanding of the pathways that control immune responsiveness is still far from complete and underscore the requirement for comprehensive preclinical testing of vaccine adjuvants before clinical use.

Vaccination is arguably the most successful medical intervention conceived to date; pathogens such as polio and smallpox virus, equated in our parents' generation with suffering and death, have been all but forgotten by our children. However, effective prophylactic vaccines against major human pathogens such as malaria, mycobacterium tuberculosis and HIV are still lacking. Likewise, therapeutic vaccines that can promote immune responsiveness in case of chronic infections or cancer have not made a substantial impact on human health to date.

It is generally expected that the development of effective vaccination strategies in such highly demanding settings will be made possible by our improved understanding of the molecular interactions that control the induction of immune responses. Analogous to the rational design of small molecule drugs, such rationally designed vaccines should target specific receptors on lymphocytes or on the antigen-presenting cells (APCs) that activate them.

The utility of vaccination is readily explained by the way our adaptive immune system detects foreign pathogens. Rather than having a preformed repertoire of lymphocytes with receptors that are specific for different pathogens, our body produces a highly diverse and random collection of lymphocytes that have the capacity to recognize essentially any potential foreign antigen. The value of this system is hard to overestimate, as it allows the body to detect pathogens that evolve rapidly. But the ability of our adaptive immune system to recognize such an immense variety of antigens has one less pleasant consequence: the frequency of lymphocytes that recognize a given antigen is extremely low.

This is where vaccination comes in. Vaccination molds our adaptive immune system by increasing the frequency and activity of those lymphocytes that we consider useful. This requires two components: the antigen against which one wishes to increase reactivity, plus an adjuvant that conveys a sense of danger.

Traditionally, such adjuvants were natural products for which the immune-stimulating effects were established empirically. Now our increased knowledge of the cellular interactions that control immune responses makes it possible to target specific pathways. For instance, this targeting may involve administration of ligands for activating receptors, such as CD40 or Toll-like receptors, present on antigen-presenting cells, or administration of cytokines such as interleukin (IL)-12 or IL-15.

Of these targeted adjuvants, the administration of ligands for CD40 has been considered particularly attractive, as previous work has demonstrated that signaling through CD40 forms a crucial switch in the induction of B-cell responses and cytotoxic T-cell responses^3,4. On the basis of these data, a first set of clinical trials using either recombinant CD40 ligand or antibodies to CD40 has been performed⁵. In addition, small molecule ligands for CD40 are currently in development⁶.

Berner et al.¹ now provide evidence for an unexpected and highly unwelcome long-term effect of this targeted vaccine adjuvant. They show that vaccination of mice with antibody to CD40, plus IL-2 (the latter for maximal T-cell expansion) can actually interfere with the long-term capacity for T-cell reactivity. Furthermore, the authors suggest that the same deleterious effect may occur with other targeted vaccination strategies that, like antibody to CD40, share a capacity to induce an interferon (IFN)-γ-dependent apoptosis of helper T cells.

The authors build their case on three pieces of data in a mouse tumor model. First, whereas administration of a tumor vaccine that contained IL-2 and antibody to CD40 enhanced tumor recognition in the short term, it reduced reactivity at later time points—even below the level observed in mice that had not been vaccinated. This finding suggests that rather than promoting recognition of tumor-associated antigens, vaccination actually leads to some level of tolerance.

Second, administration of CD40 antibody and IL-2 resulted in substantial apoptosis of CD4⁺ (helper) T cells in the days following vaccination.

Third, both the impaired tumor recognition and the increased CD4⁺ T-cell apoptosis induced by CD40 antibody and IL-2 administration were not observed in animals in which IFN-γ signaling was disrupted. These data fit well with a series of studies that show that IFN-γ production during natural infections contributes to the contraction of both antigen-specific CD4⁺ and CD8⁺ T-cell responses^7,8. But whereas this earlier work suggested that IFN-γ signaling could be important to convert a markedly high T-cell response to a physiological memory T-cell response, the current work suggests that vaccine adjuvants such as antibodies to CD40 that induce very strong IFN-γ signaling can actually deplete the memory T-cell compartment.

Notably, the mechanism through which IFN-γ signaling interferes with long-term T-cell responsiveness remains largely obscure. Furthermore, although the authors showed a correlation between CD4 T-cell apoptosis and lack of tumor recognition in animals treated with CD40 antibody and IL-2, formal evidence for a cause-effect relationship is lacking. To further resolve this issue, it will be important to determine to what extent vaccinated animals have lost CD4⁺ T-cell reactivity toward vaccine-encoded antigens and whether resupply of such reactivity can rescue the ability to control tumor growth.

In addition, it may be useful to extend this analysis to the tumor-specific CD8⁺ T-cell repertoire. Specifically, whereas the authors demonstrate that treatment with CD40 antibody has no detrimental effect on total CD8⁺ counts, it clearly remains possible that a selective loss of the tumor-specific CD8⁺ T cells that are activated by vaccination does occur, and such a loss could play a role in the subsequent immune failure. Interestingly, this latter model would fit well with the work of Bartholdy et al.², which describes a marked loss of virus-specific CD8⁺ T cells upon treatment with antibody to CD40.

Why was this deleterious effect of CD40 antibody as a vaccine adjuvant not noted before? All too frequently, researchers lack the stamina for long-term analyses and this may lead to an under-reporting of long-term side effects. Perhaps more importantly, adjuvant CD40 antibody treatment may have different effects depending on the antigen or pathogen involved. In support of this idea, whereas CD40 antibody treatment is highly detrimental in chronic lymphocytic choriomeningitis virus infection, this negative effect does not occur in certain other models of viral infection².

Given such caveats, how worried should we be? Exhaustion of long-term T-cell responsiveness may perhaps be most worrisome in clinical settings where pre-existing T-cell reactivity contributes to disease control. Specifically, naturally occurring T-cell responses are not considered a major factor in limiting the rate of progression for most human cancers. A vaccine approach that would inadvertently destroy a low-level pre-existing immune response may, in that case, not be beneficial; however, it would also not hasten disease progression. In contrast, in the case of chronic viral infections and virus-induced tumors, a delicate balance can exist between viral or tumor load and immune reactivity. In such cases, a vaccine that would accidentally reduce T-cell reactivity may well induce rapid disease progression.

The data from Berner et al.¹ and Bartholdy et al.² add a cautionary note to the clinical use of CD40 ligands as vaccine adjuvants (Fig. 1). On a more general note, the current data highlight how interventions that target the interactions of immune cells can sometimes have dramatic and unintended effects—as also evidenced by the ill-fated CD28 antibody trial in the UK last year⁹. It has recently been proposed that, because of their potency and pharmacokinetics, new guidelines may be required to assess the risk of this class of compounds before human administration¹⁰. Clearly, the type of studies reported here by Berner et al.¹ form an essential part of such analyses.

Figure 1: Pros and cons of artificial CD40 triggers.

Katie Ris

(a) APC activation upon encounter with an antigen-specific CD4⁺ T cell. Interactions that control the outcome of CD4⁺ T cell–APC encounter are exemplified by—but by no means limited to—the receptor-ligand pairs indicated. CD40L, CD40 ligand; TCR, T-cell antigen receptor; CTLA4, cytotoxic T lymphocyte–associated protein 4; MHCII, major histocompatibility complex type II protein. (b) APC activation upon triggering with artificial CD40 ligands, such as an antibody to CD40 (anti-CD40). The use of artificial CD40 ligands obviates the need for antigen-specific CD4⁺ T-cell help and can overcome tolerance in mouse model systems^3,11. However, artificial CD40 triggering has been associated with polyclonal B-cell activation and may induce auto-aggression by self-reactive T cells¹². In addition, the data from Berner et al.¹ and Bartholdy et al.² suggest that—at least under some conditions—T-cell memory formation may be impaired.