Calls to develop a universal influenza vaccine have increased in the wake of the 2009 H1N1 influenza pandemic. This demand comes at a time when analyses of the human antibody repertoire, informed by structures of complexes between broadly neutralizing antibodies and influenza hemagglutinin, have revealed the target of a class of broadly neutralizing antibodies. Recent studies suggest a path forward to more broadly protective influenza vaccines.
The unexpected geographic origin and rapid spread of the 2009 H1N1 swine origin pandemic virus thwarted efforts to immunize populations in the early days of the pandemic and underscore the importance of a universal influenza vaccine. The 2009 pandemic has also provided clues that bring us closer to this goal. This has been the first influenza pandemic during which new tools have allowed dissection of the human immune response to influenza at the molecular level. We can now obtain monoclonal antibodies from infected or immunized humans and can map the epitopes recognized by these antibodies at atomic resolution. Recent experiments using these tools, combined with more conventional analyses, inform our understanding of the evolutionary patterns that lead to influenza pandemics and suggest vaccine strategies to combat pathogens, like influenza, that evolve quickly and have high serotypic diversity. It now appears that sequential infection by virus strains that share a conserved neutralizing epitope on a background of significant antigenic change may promote the production of antibodies against that conserved epitope. Such sequential exposures could promote greater breadth of immunity.
To bind and enter cells, viruses must expose attachment- and entry-mediating proteins on their surface. These proteins are targets of neutralizing antibody responses. Type A and B influenza viruses have two chief protective antigens: hemagglutinin (HA) and neuraminidase (NA). HA mediates attachment by binding cell surface sialosides and mediates fusion of viral and cellular membranes. HA is the primary determinant of virus neutralization. In human hosts, HA evolves rapidly under selective pressure to evade immunity. NA cleaves the cell surface sialosides to which HA binds. Its activity is required for the release of newly budded virus from the infected cell surface. Immunity to NA can lessen disease but does not appear sufficient to prevent pandemic spread (Monto and Kendal, 1973).
The recent pandemic provides a new example of how immune pressure drives influenza evolution. Wei et al. (2010a) demonstrated that the HAs of the 1918 and 2009 H1 pandemic strains are antigenically similar, sharing dominant neutralizing epitopes that differ from those of recent seasonal influenza strains. Mapping historical HA sequences onto a common H1 HA scaffold showed that, as the 1918 pandemic H1 influenza circulated in humans and became a seasonal strain, new glycosylation sites accumulated on the variable HA head. These new glycans sterically blocked antibody binding to dominant neutralizing epitopes, shifting the primary targets of antibody recognition to other parts of HA. After the addition of a new glycan masked a patch of HA surface, drift in that patch slowed.
Antigenic drift, supplemented by the addition of glycans, also lays the groundwork for the re-introduction of zoonotic new pandemic influenza strains that resemble those responsible for previous pandemics. Strains with HA resembling that of the 1918 strain, which infected both swine and humans, were maintained in swine in a nearly undisturbed state, possibly due to weak host selective pressure. The 2009 zoonotic re-introduction of an old H1 antigenic variant was possible because the dominant neutralizing response to drifted and more heavily glycosylated seasonal H1 viruses had shifted to neutralizing epitopes not present in the pandemic strain. Thus, the immunity of the human herd to viruses that resemble 1918 strains had waned. Those exposed to the 1918 virus or the antigenically similar strains that circulated in the years shortly after 1918 were largely protected from the 2009 pandemic. Approximately one-third of those over 60 years of age had pre-existing antibodies that neutralized the 2009 pandemic virus, and older people were relatively spared in the 2009 pandemic (Hancock et al., 2009; MMWR, 2010). Although young people lacked significant titers of pre-existing neutralizing antibodies, those old enough to have experienced seasonal H1 influenza infection or immunization responded rapidly to a single dose of un-adjuvanted influenza vaccine, suggesting that they had been primed by their prior exposures (Clark et al., 2009).
The existence of such priming is confirmed by experimental studies in animals (del Guidice et al., 2009). Immunization of naive ferrets with seasonal influenza vaccine does not elicit detectable neutralizing or hemagglutination inhibiting (HI) titers against the 2009 pandemic strain. However, seasonal immunization primes ferrets for higher neutralizing and HI titers upon immunization with a single dose of adjuvanted pandemic H1N1 vaccine. If the seasonal prime and pandemic boost both are adjuvanted, ferrets are protected even from nasal shedding of a pandemic challenge virus. What is the mechanism of priming by the seasonal vaccine? The seasonal vaccine could elicit cross-reactive CD4+ T cells that subsequently help to generate affinity matured antibody responses to the pandemic vaccine. Alternatively, despite its inability to elicit significant neutralizing titers to the pandemic strain, the seasonal vaccine could generate memory B cells that produce cross-reactive antibodies that are not detected in serum neutralization tests against pandemic H1N1 because they have low affinity for the pandemic strain or are rare. Subsequent immunization with the pandemic vaccine could selectively expand the cross-reactive memory B cells and induce them to produce affinity matured antibodies that better neutralize the pandemic strain.
The structural basis for cross-neutralizing antibodies is increasingly well understood. Such antibodies have been elicited in mice (Okuno et al., 1993) and now have been detected by cloning the B cell repertoires of immunized and infected individuals. Some functional antibodies that cross-react within a subtype bind the head of HA, but structural studies show that the most broadly neutralizing antibodies, those that can cross-neutralize between subtypes, bind a highly conserved stem epitope (Ekiert et al., 2009; Sui et al., 2009). Indeed, this epitope could be the ‘Achilles heel’ of influenza virus. This epitope is present in all influenza vaccines. Yet, current immunization regimens appear to be relatively ineffective at eliciting antibodies that recognize it. Instead, potent but less broadly neutralizing antibodies against variable epitopes on the HA head dominate after conventional immunization.
A recent study by Wei et al. (2010b) demonstrated that a non-traditional immunization regime can elicit cross-reactive antibodies recognizing this epitope in previously influenza-naive experimental animals. Priming with injections of plasmid DNA encoding a seasonal H1 HA followed by boosting with a seasonal subunit vaccine or a replication-defective adenovirus vector encoding seasonal H1 HA elicited a broadened pseudotype neutralization response within the H1 subtype and some cross-reactivity against H2N2 and H5N1 viruses. Two doses of non-adjuvanted subunit vaccine also broadened the response within the H1 lineage but to a lesser extent than the four-dose DNA prime/subunit or vectored boost regime. The increased breadth of neutralization could be due to the larger number of doses of vaccine or to the prime-boost design of the non-traditional regimen. To test whether the increased breadth was mediated by antibodies recognizing the conserved stem region of HA, they generated a mutant H1 HA in which a glycan masks the broadly neutralizing stem epitope. By preabsorbing sera with cells expressing this mutant HA or a wild-type HA (to differentially remove antibodies recognizing the stem epitope), they determined that responses to the stem epitope were indeed responsible for the observed breadth of neutralization. This finding establishes that it is possible to elicit broadly neutralizing antibodies recognizing the stem epitope with good efficiency in animals with no previous influenza exposures.
Can immunization or infection elicit such antibodies efficiently in humans, especially in people whose immune response to influenza is influenced by prior exposures? A new study (Wrammert et al., 2011) presents the promising finding that natural infection of adults with the 2009 pandemic strain can elicit such antibodies. They cloned antibodies from plasmablasts of patients who were ill with 2009 H1N1 pandemic influenza. Plasmablasts are antibody-producing cells that expand in the early days after an antigenic challenge. Most antibodies that recognized the pandemic H1 HA also recognized seasonal H1 HA. Using single-cell PCR, they generated monoclonal antibodies from 15 HA-specific plasmablasts. A number of the antibodies cross-neutralized seasonal and pandemic H1 strains and bound H5 HA. The most cross-reactive of the antibodies compete with a known stem-binding antibody (C179), indicating that pandemic H1 influenza infection of individuals who presumably had previous exposures to seasonal H1 influenza elicited broadly neutralizing antibodies against the stem epitope.
Examination of relative binding affinities and degree of somatic hypermutation of the antibodies provided further insights into their origin. Antibodies that recognized both seasonal and pandemic H1 HA, but preferentially bound the pandemic strain, had the highest number of somatic mutations, suggesting that they arose from re-stimulation and affinity maturation of memory B cells that were first generated by past seasonal exposures. Antibodies that only recognized the pandemic strain had the least number of somatic mutations, suggesting that they arose from naive B cell clones. These results suggest that broadly neutralizing antibodies can be raised naturally by sequential infections with seasonal and pandemic H1 strains.
Why did the sequential antigenic exposure (seasonal followed by 2009 pandemic H1N1) raise broadly neutralizing antibodies against the stem epitope? This epitope is conserved between the 2009 pandemic H1 HA (or the 1918 HA) and seasonal H1 HA (Figure 1). Therefore, any memory B cells elicited against this region by an old seasonal infection are likely to be preferentially stimulated to proliferate and affinity mature upon subsequent pandemic H1 influenza exposure. The greater number of shared epitopes between two sequential seasonal H1 strains would not favor a preferential response to this region to the same extent following two consecutive seasonal infections.
Wrammert et al. leave some important questions for further study. Are the polyclonal sera of patients infected with the 2009 pandemic strain generally more broadly neutralizing than sera elicited by other influenza infections? The relatively high proportion of broadly neutralizing antibodies was observed in the plasmablast repertoire, the repertoire of the acute response to infection. Will these antibodies still be present in high proportion after the response has matured and become established in memory? Would sequential immunization with H1 seasonal and H1 pandemic live attenuated vaccines, subunit vaccines or adjuvanted subunit vaccines be as effective as these patients’ exposure histories, which culminated with an episode of pandemic influenza illness, at eliciting such antibodies? Nevertheless, the hypothesis that serial exposures to antigenic variants that share a conserved epitope will preferentially elicit a response against the conserved epitope is plausible and provides a potential technique to focus the immune response on an epitope of interest. If successful, this technique could also be generalized for immunization against other antigenically variable pathogens.
In 2009, nature re-introduced a historical antigen through a new zoonosis. We can emulate this sequence with an immunization regimen. The growing library of sequence information makes it possible to pick immunogens from across time and geography, creating exposure sequences that could not occur naturally. This freedom could allow immunization regimens that re-program the immune response and increase the breadth of elicited immunity against influenza and other pathogens.