Vaccines are currently the best countermeasure against influenza virus infections in humans. However, available vaccines are limited in breadth such that they are effective against only the circulating strains that are closely related to the vaccine strain. The genetic variation of influenza viruses allows them to escape immune responses quickly, a mechanism called antigenic drift. Therefore, influenza vaccination programs require annual revaccination with reformulated vaccines—a cumbersome and expensive undertaking. Because vaccine strain selection is based on surveillance and prediction, mismatches between vaccine strains and circulating viruses occur, resulting in a sharp drop of vaccine efficacy1. Furthermore, the current vaccine appears to be less effective in the elderly2. The emergence of the pandemic H1N1 virus in Mexico in 2009 and the more than 140 human cases of H7N9 virus infections in China3 earlier this year remind us of the constant threat of zoonotic influenza viruses (Fig. 1). Seasonal influenza virus vaccines do not protect from infection with novel pandemic viruses, and it takes at least 6–8 months to develop, test and produce conventional vaccines against emerging viruses. For example, the first batches of pandemic vaccine in 2009 were shipped for distribution after the first pandemic wave hit the population and were therefore too late4. It should be noted that many pre-pandemic vaccine studies with avian H5 or H7 vaccines reported very low vaccine efficacy in humans, and some of these vaccines might not protect from morbidity even if they were available in time5.

Figure 1: Schematic of influenza virus strains circulating in the human population and zoonotic influenza viruses that sporadically cause disease in humans.
figure 1

H1N1, H3N2 and influenza B strains of both lineages co-circulate in humans. The seasonal H1N1 strain was replaced in 2009 by the pandemic H1N1 strain, which now circulates in the population. Current influenza virus vaccines are quadrivalent, containing isolates from all four types of circulating viruses. Human influenza viruses undergo constant change (antigenic drift, indicated by the changing colors of the bars), and vaccines therefore have to be reformulated annually in a cumbersome and expensive process. Zoonotic influenza viruses of the H5 and H7 subtypes infect humans occasionally. Years in which such infections were reported are indicated by dark gray (H5) and light gray (H7) bars. Most recently an avian H7N9 virus caused more than 140 human infections in China. Seasonal influenza virus vaccines do not protect against these and other potential pandemic viruses.

Full size image

Targeting conserved regions of the virus

To overcome the shortcomings of seasonal influenza virus vaccines and to enhance our pandemic preparedness, game-changing influenza virus vaccines that confer broad—ideally universal—and long-lasting protection are needed. It is desirable that such a vaccine would even induce lifelong immunity after administration of a vaccination regimen comparable to the ones used for polio, measles and other viral diseases. Several promising approaches have been developed recently, including antibody- and T cell–based strategies6,7,8,9,10,11. The majority of these approaches target conserved regions of hemagglutinin (HA), the major surface glycoprotein of the influenza virus. Most of the humoral immune response against influenza viruses is directed against the HA12; this response drives antigenic drift in the membrane distal globular head region of the HA molecule, which exhibits high plasticity. However, this plasticity is constrained by the two functions of HA—mediation of the binding of the virus to host cell-surface receptors (through the receptor binding site) and of the fusion between viral and endosomal membranes (through the stalk domain). To stay viable, the virus has to retain the structure of both the binding site and the stalk domain. These regions are therefore conserved in the influenza virus HA. The high conservation of these structures is the weak point of the virus and can be targeted by broadly neutralizing antibodies and universal vaccines.

Vaccination strategies based on conserved parts of the HA, mainly the stalk domain (Fig. 2), have been tested successfully in mouse, ferret and nonhuman primate models6,7,8,9,10. However, these vaccination strategies are dependent on—or at least influenced by—pre-existing immunity. Humans are exposed multiple times to influenza viruses by clinical and subclinical infection as well as by vaccination throughout their lifetimes, resulting in a very complex history that shapes the immune response to new influenza viruses or vaccines. Vaccines that target conserved regions of the influenza virus (such as the stalk domain of the HA) can be based on boosting low levels of pre-existing immunity against these regions. However, it is almost impossible to mimic the human situation in animal models. The only way to explore whether these vaccines are effective in humans is to test them in clinical trials. To progress quickly with this endeavor, it would be pragmatic to set up a number of small, exploratory trials that compare different approaches (such as chimeric HAs9,10 or heterologous prime-boost regimens7). Vaccine candidates could be produced through different vaccine platforms (such as inactivated split and whole-virus vaccines, live attenuated vaccines, recombinant protein vaccines or virus-like particles). These different vaccine platforms might be suitable for universal influenza virus vaccine candidates.

Figure 2: Universal influenza virus vaccine strategy based on chimeric HAs.
figure 2

(a) Example of a chimeric HA: cH6/1 is a combination of a conserved stalk domain from H1 HA (in blue) and an 'exotic' globular head domain (in red) from H6 HA. (b) Human vaccination regimen based on chimeric HAs. Most individuals already have pre-existing immunity against full-length H1 HA (in blue), including low levels of B cells with specificities for portions of the conserved stalk domain. These stalk-specific antibodies might then be boosted by sequential vaccination with, for example, cH5/1 (H5 head on top of an H1 stalk) and cH6/1 (H6 head on top of an H1 stalk) HAs. Humans are naive to the exotic H5 (green) and H6 heads (red) and therefore mount primary immune reponses only against these (usually immunodominant) structures. The conserved stalk domain (blue) is repeatedly presented to the immune system, and antibodies against this (usually immunosubdominant) structure are boosted. Structures are based on PDB 2WRG. All HA structures are shown as monomers.

Full size image

It is important to make sure that as much data as possible are collected during these trials to obtain comprehensive results that will provide the basis for further development of a down-selected set of successful universal vaccine candidates. Because most recent universal vaccine approaches9,10 do not rely on the hemagglutination inhibition titer, the classical correlate of protection, this opportunity could also be used to validate novel correlates of protection. Broadly neutralizing antibodies are known to work through a variety of mechanisms that include direct neutralization by inhibition of fusion of viral and endosomal membranes13,14,15,16,17 and inhibition of viral egress or HA maturation13,14,15,16. Indirect mechanisms such as complement-dependent lysis of virus or cells18 and antibody-dependent cell-mediated (ADCC) cytotoxicity might have important roles as well. Recently developed methods to measure these activities, include entry inhibition assays, modified micro-neutralization assays that capture events downstream of entry, ELISA to measure stalk-reactive antibodies19,20,21 and the use of Fcγ receptor–humanized mice for passive-transfer experiments. Additionally, the vaccine response needs to be evaluated on the level of B and T cells. The plasmablast response in particular can give extraordinary insights into diversity and specificity of vaccine induced humoral immune responses on a monoclonal level12. Small challenge studies in humans could be used to validate these correlates of protection.

In addition to comparing different universal vaccine approaches and vaccine platforms, it is also desirable to compare adjuvants that have been developed for influenza virus vaccines so far. The use of certain adjuvants is known to prolong substantially the half-life of immune responses—a desired feature for a universal vaccine. The use of adjuvants in seasonal influenza virus vaccines is seen critically by regulatory agencies and the public, especially in Europe. However, adjuvants would be used for only a limited number of vaccinations—probably three or fewer—in a long-lasting universal influenza virus vaccine, making complications less likely. Epitopes that are thought to be associated with diseases such as Guillain–Barré syndrome or narcolepsy could be eliminated from vaccine constructs to minimize possible risks. Recent reports about vaccine-associated enhanced respiratory disease (VAERD), sometimes caused by non-neutralizing cross-reactive antibodies induced by mismatched whole inactivated virus- or influenza matrix protein-2 ectodomain (M2e)-based vaccines in pigs, are of potential concern22,23,24. An early indication that a similar phenomenon might occur in humans25 has been questioned by many other studies26,27,28,29. Furthermore, current approaches to universal influenza virus vaccines work through neutralizing antibodies and are therefore unlikely to cause a similar disease, even in pigs.

In contrast to adults with pre-existing immunity, young children are naive to influenza virus. It might therefore be necessary to prime children before putting them on a regular universal influenza virus vaccine regimen. This priming could be done using either regular inactivated virus vaccines or with regular live-attenuated vaccine, which shows good and broad efficacy in this cohort. Primed children could then receive one or two boosters with a universal influenza virus vaccine and might be protected for life. The situation for the elderly is completely different. This cohort might have good pre-existing immunity but has the tendency to react poorly to vaccination owing to immune senescence. However, an individual who is vaccinated at a younger age with a vaccine that induces high and long-lasting titers of broadly neutralizing antibodies might be protected even at old age. This is conceivable, as examples of long-lasting immunity against influenza viruses exist30, although these long-lasting responses are usually induced by viral infection. However, a recent study showed that high titers of stalk-reactive antibodies induced by the New Jersey 1976 H1N1 swine influenza virus vaccine were still detectible after more than 30 years31. Again, the use of adjuvants in combination with the right vaccine platform might potently enhance the half-life of broadly neutralizing antibodies.

The initial cost of conducting these trials might seem high, but investment in universal influenza virus vaccine approaches might make it possible to overcome the threat of seasonal and pandemic influenza. It is conceivable that such an approach would lead to a drastic reduction of influenza viruses circulating in the human population21,32. Influenza B viruses, which have no animal reservoir, might be eradicated completely. Furthermore, important lessons could be learned for the fight against other genetically variable viruses, such as HIV and hepatitis C, which could lead to the development of groundbreaking new vaccine technology against these and other viral pathogens.