When it comes to protection from influenza infection, antibodies are best. Neutralizing antibodies to the surface glycoproteins of influenza type A and B viruses, hemagglutinin and neuraminidase, block binding to receptors on the respiratory epithelium, preventing virus uptake and subsequent release of new virus particles (Fig. 1). But influenza viruses move quickly: the selective pressure of the antibody response on seasonal influenza viruses drives the emergence of escape mutants that can cause epidemics in communities immune to earlier strains1. This is why seasonal influenza vaccines require frequent updating.

Figure 1: Role of virus-specific memory T cells in protection from influenza virus challenge.
figure 1

During primary infection of the respiratory tract, influenza viruses bind receptors on the epithelial cell surface via their surface hemagglutinin and are internalized (left). The eight segments of the viral RNA genome are transcribed and translated to produce viral proteins, including the highly variable hemagglutinin and neuraminidase and the more conserved internal proteins, enabling the production of new virions. The host produces antibodies mainly directed against hemagglutinin and neuraminidase that neutralize the virus and prevent further infection. Intracellular degradation of newly synthesized or endocytosed viral proteins produces peptide fragments, some of which (T cell epitopes) bind host class I or class II major histocompatibility complex (MHC) molecules and can be recognized by class I–restricted CD8+ or class II–restricted CD4+ effector T cells bearing T cell receptors (TCRs) of appropriate specificity. These effector T cells contribute to viral clearance, for example, by lysing infected cells and secreting soluble mediators. On challenge with a variant (seasonal or pandemic) influenza virus that is not neutralized by pre-existing antibodies, memory T cells specific for conserved viral epitopes shared between the priming and challenge viruses are recruited to the site of infection (right). The data of Wilkinson et al.3 suggest that these pre-existing CD4+ T cells can reduce virus shedding and the severity of infection, possibly by direct lysis of infected epithelial cells.

The specificity of the antibody response also underlies the emergence of influenza pandemics. Pandemics occur when gene shuffling between animal and human influenza A viruses produces a human-transmissible virus with a new hemagglutinin against which most of the population lacks immunity. This has happened four times in the last century. On each occasion, antibodies against contemporary seasonal viruses failed to protect against the new virus subtype: A(H1N1), A(H2N2), A(H3N2) and a new A(H1N1) in 1918, 1957, 1968 and 2009, respectively.

The unexpected nature of the 2009 A(H1N1) pandemic and the recent human-to-human transmission of a variant A(H3N2) virus of swine origin in the US2 remind us that new influenza viruses are a continuing and unpredictable threat. In an ideal world, vaccines would induce durable broad-based immunity that would minimize the impact of these viruses. In this issue of Nature Medicine, Wilkinson et al.3 present evidence that CD4+ T cells activated by earlier influenza virus infections reduced the severity of disease when seronegative volunteers were challenged with seasonal influenza viruses. These observations point to CD4+ T cells as potential targets for activation by vaccines that would protect not only against new seasonal influenza variants but also against future pandemic viruses.

There is evidence that T cells can mediate cross-protective (heterosubtypic) immunity. In classic studies, it was shown that adults who had previously experienced an A(H1N1) infection were protected from the pandemic A(H2N2) virus when it emerged in 1957 (ref. 4) and that specific cytotoxic T lymphocyte activity was associated with low virus shedding in infected seronegative volunteers5. T cells cannot prevent virus infection but they can sense infected cells by recognizing fragments of viral protein (epitopes) complexed to human leukocyte antigen (HLA) molecules on the surface of infected epithelial cells or antigen-presenting cells (Fig. 1). As T cells preferentially see epitopes derived from the conserved internal proteins of the influenza virus, cross-protective immunity has been attributed to pre-existing cytotoxic CD8+ T cells that kill virus-infected cells presenting these conserved epitopes. In this way, CD8+ T cells might reduce the duration and severity of infection with a seasonal variant or pandemic virus that escapes the protective antibody barrier. Indeed, many of the CD8+ T cell epitopes identified in internal proteins of A(H3N2) or pre-2009 A(H1N1) influenza viruses are shared with the highly pathogenic avian A(H5N1), the 2009 pandemic (A(H1N1)pdm09) or both influenza viruses, and CD8+ T cells specific for some of these epitopes have been detected in the blood of healthy donors6,7,8.

Cross-protection by CD4+ T cells has received less attention, perhaps because mouse studies have suggested they function mainly to promote antibody responses9. Nevertheless, conserved CD4+ T cell epitopes have been identified in internal and surface proteins of human and avian influenza viruses. Some of these epitopes, at least, are naturally generated in infected cells, and CD4+ T cells from seronegative donors can respond to them7,8,10,11.

Wilkinson et al.3 have now shown that virus-specific memory CD4+ T cell numbers predict the outcome of human influenza infection. They counted virus-specific T cells in the blood of volunteers before and after challenge with recent seasonal viruses of the A(H3N2) subtype or the former seasonal A(H1N1) subtype, which circulated until it was replaced by A(H1N1)pdm09 in 2009. Their key finding was that pre-existing virus-specific T cell numbers were inversely related to the severity of illness after challenge3. Remarkably, it was the CD4+, rather than the CD8+, T cell counts that correlated with lower virus shedding and shorter illness in response to A(H3N2) challenge and with lower symptom scores in response to both challenge viruses. By contrast, virus-specific T cell numbers after challenge correlated positively with viral load and symptoms3.

The presence of pre-existing virus-specific T cells was not surprising, as the subjects were old enough to have experienced several influenza infections (range 18–45 years). Subjects lacked detectable baseline serum antibodies to the challenge virus, suggesting that they had not recently been exposed to this subtype. As antibodies were not measured to the alternative subtype or the A(H1N1)pdm09 virus (which emerged shortly before the A(H1N1) challenge), T cell numbers might have been boosted by recent exposure to one of these viruses.

If memory CD4+ T cells are responsible for the reduced disease severity, how do they exert their effects? There are several possibilities. Given that they are primed and elevated in number, they might accelerate the development of specific antibody and/or CD8+ T cell responses, though perhaps not quickly enough to ameliorate symptoms in the first few days after challenge. Memory CD4+ T cells might produce or induce mediators that enhance recruitment and activation of innate cells that contribute to viral clearance12. Alternatively, they might exert direct cytotoxic activity against virus-infected cells. The authors favor the last possibility, as they found that memory CD4+ T cells showed specific granule-mediated cytotoxicity in vitro3. Whether this activity is relevant in vivo is not yet known and will be difficult to evaluate directly in humans.

Whatever their mechanism of action, the relevant CD4+ T cells must either reside in or be rapidly recruited to sites of viral antigen presentation in the respiratory tract or lymphoid tissue. They must also recognize viral peptides presented with class II HLA molecules at those sites, either on respiratory epithelium or on other antigen-presenting cells. Although neither of these requirements was assessed in the infected subjects, the authors showed expression of the class II molecule HLA-DR in explanted normal lung and cultured primary bronchial epithelial cells; expression was elevated in the latter cells after influenza virus infection in vitro3.

Do these new data mean that pre-existing cross-reactive CD8+ T cells do not contribute to influenza immunity? It would be premature to draw this conclusion for several reasons. For example, the study may have lacked power to test this possibility, as the number of subjects in this study was small. Additionally, cross-protective cells may be a minority of virus-specific memory CD8+ T cells, too few to influence the relationship between cell number and symptoms, or they may reside predominantly in the respiratory tract rather than the circulation.

If memory CD4+ T cells do play an important part in cross-protection, how long do they persist after infection? T cell immunity declines markedly over the first 1–2 years13,14, so maintenance of cross-protection must rely on regular re-exposure to influenza virus. Conventional inactivated influenza vaccines produced by most global manufacturers protect by inducing subtype-specific antibodies to hemagglutinin and do little to boost cross-protective T cell responses. However, vaccines that use live attenuated viruses or some new antigen delivery or adjuvant technologies can activate both antibody and T cell responses and offer the prospect of some cross-protection15. These vaccines will have to prime a more durable cross-protective response than natural infection to remove the need for regular vaccination against the latest circulating viruses. The study by Wilkinson et al. suggests that activation of long-lived cross-protective CD4+ T cells should be one of the goals of future vaccine development.