The purpose of our study was to develop a better understanding of the nature of the H1N1 viruses that provided serological cross-reactivity and protection against the 2009 pH1N1 virus. We primed ferrets with H1N1 influenza viruses of variable antigenic distance isolated from the 1930s to the 1990s and determined the effect of prior infection on subsequent challenge with 2009 H1N1 pandemic influenza virus. We postulated that H1N1 viruses antigenically related to the 2009 pH1N1 virus would protect against challenge with this virus, while those that were antigenically dissimilar would be unable to protect against challenge. We found that prior infection with a similar swine origin virus (NJ/76) or viruses isolated in and prior to 1947 provided significant protection against CA/09 challenge, while significantly lower protective efficacy was observed with viruses isolated in and after 1950. Our data suggest that the addition of glycosylation sites to the HA globular head, and the specific location of those sites, contributed to this loss in antigenic cross-reactivity and protective efficacy.
Soon after the pH1N1 virus appeared, the Sa antigenic site of the CA/09 HA was shown to be similar to that of the 1918 virus. Manicassamy et al. generated inactivated influenza vaccines or VLPs from a panel of 11 H1N1 viruses isolated between 1918 and 2009 and evaluated the ability of these vaccines to protect mice from subsequent challenge with a mouse-lethal 2009 pH1N1 virus (39
). 1918 VLPs, inactivated Weiss/43, and inactivated classical swine H1N1 viruses protected mice from lethal infection with the pH1N1 virus, while 1977, 1984, 1991, and 2007 H1N1 viruses offered only partial protection (39
). The authors of that study generated and sequenced escape mutants of CA/09 using three monoclonal antibodies that cross-reacted with the 1918 and 2009 H1N1 HAs, confirming that the Sa antigenic site was conserved between the 2009 pH1N1 and 1918 viruses. They concluded that prior exposure to human influenza viruses from 1918–1943, or vaccination with classical swine H1N1 virus (NJ/76), offered significant levels of cross protection against the 2009 pH1N1 virus, thus explaining the lower incidence of disease and infection in persons aged 65 years or older (39
). Although Weiss/43-vaccinated mice lost weight after challenge, they recovered; these mice did not have cross-reactive HAI antibodies against the CA/09 virus. Our findings confirm those of Manicassamy et al. and extend them by demonstrating the ability of 1935 and 1947 but not 1950 H1N1 viruses to protect ferrets from infection with the CA/09 virus.
From their determination and analysis of the crystal structure of the CA/09 HA, Xu et al. concluded that the 1918 HA was a remarkably close relative of the CA/09 HA, with only 20% amino acid difference in the antigenic sites, that was mainly restricted to the Ca region, while the CA/09 virus differed substantially from seasonal H1 HAs in all four antigenic regions (62
). These researchers also analyzed the presence of glycosylation sites on the HA that can mask the protein surface from antibody recognition and reported that the CA/09 HA, like 1918 HA, does not have glycosylation sites in the Sa region of the globular head, while human H1 HAs have gradually acquired up to three such sites in the relatively conserved Sa region from 1930 to 2007. The lack of glycosylation of the Sa site in the CA/09 HA thus leaves the epitope exposed for antibody recognition, and the authors proposed that Sa-specific antibodies were potentially the major underlying basis of age-related immunity to the 2009 H1N1 virus. They mapped the binding on the 1918 HA of a monoclonal antibody (2D1, isolated from a survivor of the 1918 pandemic) that cross-reacted with the CA/09 HA and found that the footprint of the Fab on the HAs largely coincided with and extended the Sa region and that this site was well conserved in the 1918 and 2009 but not in seasonal H1 HAs. Interestingly, 7 amino acid differences, including three in the Sa site, along with two potential glycosylation sites in the center of the epitope (residues 129 and 163 in the Xu paper correspond to residues 142 and 177 in our paper), explain why 2D1 does not cross-react with seasonal H1 viruses (62
). Our findings with the pseudovirion assay confirm these observations of Xu et al. We have also extended their observations by demonstrating antibody cross-reactivity and cross-protection induced by prior infection in vivo
A correlation was observed between the cross-reactivity of antibodies induced by the priming viruses and the protective efficacy against the CA/09 virus. NJ/76, which provided complete protection from CA/09 challenge, was the only H1N1 virus used in the present study that was antigenically related to the CA/09 virus based on HAI and neutralization assays. This is most likely because both are swine origin viruses and both derived their HA from a classical swine virus (31
). Our results are consistent with previous studies from our lab and others that have shown that prior exposure to the NJ/76 virus or similar 1976 swine viruses, or prior vaccination with the 1976 “swine flu” vaccine, can protect against 2009 pH1N1 virus infection (31
). Individuals who received the 1976 swine flu vaccine had neutralizing antibodies against the 2009 pH1N1 virus (42
). Prior infection with AK/35 and FM/47, which provided significant, albeit partial, protection against challenge, generated lower levels of cross-reactive HAI and neutralizing antibodies against CA/09 virus. FW/50, which provided some protection against challenge, induced cross-reactive antibodies against CA/09 virus that were only detectable by ELISA. NC/99 and B/Mal, which provided no protection against challenge, did not induce cross-reactive antibodies against CA/09 virus.
Interestingly, prior infection with AK/35 and FM/47 provided similar levels of protection against CA/09 challenge. Although AK/35 is closely related to swine viruses, FM/47 is not (46
). FM/47 was described as an A prime influenza virus that emerged after a major antigenic change occurred in the H1N1 influenza virus in 1947 (34
) and would not be expected to provide similar levels of protection as AK/35. Whereas NJ/76 shares significantly more homology (91%) in the total HA protein sequence with the CA/09 virus, the other H1N1 viruses share similar levels of homology: AK/35 (81%), FM/47 (81%), FW/50 (80%), and NC/99 (79%). Thus, homology of the complete protein sequence does not explain the differences in protective efficacy. However, AK/35 and FM/47 share significantly more homology with CA/09 in the Sa antigenic site compared to FW/50, 62% compared to 46%, which may contribute to the differences observed in the protective efficacy of these viruses.
A significant loss in protective efficacy was observed following primary infection with the viruses isolated in 1947 (FM/47) and 1950 (FW/50). Although some studies have shown that viruses isolated prior to 1950 can provide at least partial protection against CA/09 virus challenge, only one H1N1 virus isolated from 1948 to 1957 (Denver/1/1957) has been tested, and this virus did not offer significant protection from CA/09 challenge (31
). Our study now provides a much more specific time frame (between 1947 and 1950) as to when H1N1 viruses evolved to the point where they no longer offered effective protection against 2009 pH1N1 infection. This allows us to closely examine the viruses from this period to better understand why these viruses are able or unable to protect against 2009 pH1N1 infection. There was an unusually severe influenza A virus H1N1 epidemic in 1950 and 1951 associated with significant morbidity and mortality, especially in the United Kingdom and Canada; mortality rates in these countries exceeded those of the 1957 H2N2 and 1968 H3N2 pandemics (26
). Interestingly, the 1951 epidemic exhibited geographic disparities in influenza-related deaths, illustrated by high death rates in England and Canada and low death rates in the United States, except in the northeast (12
). These disparities are explained in part by laboratory surveillance reports by the World Health Organization (25
), indicating that two antigenically distinct influenza virus A/H1N1 strains cocirculated in the northern hemisphere during the 1951 epidemic: the “Scandinavian strain” isolated in northern Europe was associated with mild illness, and the “Liverpool strain” was associated with severe illness and death in Great Britain, Canada, southern Europe, and Mediterranean countries. However, it is not known whether FW/50 was isolated during this epidemic or whether it was associated with clinically severe disease, although antigenic analysis suggests that FW/50 resembled the “Scandinavian” strain (33
H1N1 influenza viruses underwent a major antigenic change in 1949 that included the addition of several new glycosylation sites that may be responsible for the changes in protective efficacy between FM/47 and FW/50 (51
). Before this time, H1N1 viruses typically had one to three glycosylation sites, but later it was common to find eight to ten glycosylation sites on the HA (51
). However, the contributions of additional mutations in H1N1 viruses in 1949, independent of glycosylation on the HA globular head, are unknown. Our study shows that the addition of glycosylation sites to the HA globular head and the location of these glycosylation sites can significantly reduce antibody cross-reactivity. FW/50 and NC/99 possess glycosylation sites at amino acids 172 and 177, respectively (60
), that are located within the Sa antigenic site on the HA (62
) (see also Fig. S2 in the supplemental material). A glycosylation site in this region likely contributed to the loss in protective efficacy shown by these two viruses compared to the other viruses.
Although our results, as well as the results from others (31
), have shown that viruses isolated prior to 1950 can protect against 2009 pH1N1 infection, it is possible that not all pre-1950 H1N1 viruses protect equally well against 2009 pH1N1 infection. Several H1N1 isolates from the 1940s, such as Hickox/40, which was shown to only partially protect against 2009 pH1N1 infection (31
), are glycosylated at amino acid 178 or 179 (60
). Prior infection with mouse-adapted FM/47 or prior vaccination with Weiss/43, neither of which express a glycosylation site at this position, provided significant protection against 2009 pH1N1 infection (39
). Thus, the protective efficacy of an H1N1 virus against 2009 pH1N1 infection depends not only on the antigenic distance from the 2009 pH1N1 virus but also on the glycosylation of the HA globular head. Unfortunately, it is difficult to make generalized conclusions from an analysis of a small number of viruses in mice and ferrets, and these studies cannot assess whether glycosylation of the HA played a role in shaping population immunity. However, this was not the intent of this study; rather, it was performed to test the hypothesis that changes in HA glycosylation affected the specificity of the immune response in subjects infected by the 2009 virus. We selected one virus from each decade of H1N1 virus circulation for the study, having considered the results from previous studies using mice that were infected or vaccinated with three viruses isolated in 1943, 1947, and 1957 (39
). Despite these limitations, data from experiments in animal models provide useful insights into serologic and epidemiologic observations from the 2009 pandemic.
Overall, we show that H1N1 viruses of variable antigenic distances from the 2009 pH1N1 virus protect to various degrees against subsequent infection with the 2009 pH1N1 virus. A significant reduction in cross-protection conferred by H1N1 viruses was observed between 1947 and 1950, which is likely due to major antigenic changes in H1N1 viruses during this time, specifically associated with the acquisition of glycosylation. This is a likely explanation for the observation that the elderly were protected against 2009 pH1N1 infection while younger adults were more susceptible.