The age group ≥ 61 years - The virtues of Antigenic Sin with the right antigens
The doctrine of Original Antigenic Sin states that an individual's first infection with an influenza A-type virus produces an indelible mark upon that person's immune system including an antibody response that persists for many years, possibly a lifetime. Though necessarily not a part of the original surmise, it is now believed that the OAS immune response includes a cytotoxic T-cell component [18
]. Based on this doctrine, persons who first encountered influenza viruses in the 1930s and early 1940s should produce immune responses upon influenza virus infection cross-reactive to similar viruses, among which is novel H1N1 (Figure ). Persons born after H1N1 viruses left circulation in 1957, including the period during which re-emergent H1N1 circulated, i.e., from 1977 onward [19
], would not be expected to produce an OAS response that would include antibody cross-reactive to novel H1N1. Indeed, the presence of antibodies to novel H1N1 [17
] and the fact that there are few clinical cases among those over 60 years of age suggests that a B-cell mediated antibody response to viruses that circulated prior to 1948 is probably present in those age ≥ 61. Figure highlights (in green) three sites for potential glycosylation all of which lie within the so-called Sa antigenic region on the globular head of H1 [21
]. The B-cell epitopes defined on the globular head of early H1N1 viruses including the 1918 strain [22
] present a much closer match by amino acid sequence to 2009 novel H1N1 than do recent circulating seasonal H1N1 viruses suggesting the mechanism of this immunity (data not shown). Interestingly, a similar response was also observed in elderly persons over age 77 in the US and Canada relative to H3N2 viruses and the pandemic of 1968/9 [23
]. We intend that one possible contribution of this paper be the suggestion that loss of antigenicity may occur by the addition of sites of glycosylation adjacent to structurally antigenic sequences in addition to the obvious restructuring of antigenic sites by direct amino acid substitution.
The age group < 32 years: Antigenic Sin with inappropriate antigens
The H1N1 virus that re-emerged into circulation in 1977 was most similar to those that circulated between 1948 and 1950. The year prior, 1947, was notable for a generalized failure of the seasonal flu vaccine [24
], and it was concluded that a very large genomic change, ascribed to intrasubtypic reassortment, had taken place between 1943 and 1947. This change involved at least five antigenic sites. Mortality associated with pneumonia and influenza declined by 2/3 between 1937 and 1947. It did not decline further between 1947 and 1957, the pandemic year in which H2N2 viruses displaced H1N1 viruses.
The progressive N-linked glycosylation of sites near those antigenic to the host is an adaptive evolutionary strategy employed by influenza viruses [25
]. The added carbohydrate molecules shield immunodominant sites permitting the adapting virus to escape preformed immunity in hosts previously exposed to less glycosylated surface proteins. As a counter-balance to this method of immune escape, however, highly glycosylated viruses are better recognized by innate defenses such as collectins, and are less virulent in the lower respiratory tract. The sequential addition of glycoconjugates to the virus during adaptation may therefore also contribute to the attenuation of later strains compared to their pandemic forebears [26
The 1918 pandemic virus had a hemagglutinin (HA) protein that contained five sites at which post-translational modification by glycosylation could occur [27
]. Of these sites for potential glycosylation, four were located in the fusion domain and highly conserved. The remaining one was in the vestigial esterase domain and appeared variably in viruses until about 1950 at which point it became evolutionarily fixed (see Figure ). Between 1918 and 1947, three new glycosylation sites appeared in circulating strains, all in areas where attachment of glycoconjugates might interfere with antibody binding. However, in most sequences of strains which circulated prior to 1949, only one to three sites were present in any single strain on the globular head. Beginning with the 1949-1950 viruses, the immediate ancestors of the H1N1 strains that re-emerged in 1977, eight to ten total potential glycosylation sites were present in these and all subsequent viruses, with four to six of these located on or near the globular head where they could interfere with antibody binding and immunity, including three within the Sa antigenic region. H3N2 viruses have gone through a similar progression of added glycosylation sites since their introduction in 1968 [25
Although glycosylation of HA can modify antibody based immunity, CD8+ T-cell epitopes have been highly conserved since the 1930s [28
], and therefore, T-cell based immunity should be unaffected by evolutionary changes in HA. Indeed, a recent analysis of both B-cell and T-cell epitopes of novel H1N1 demonstrates that only 31% of potential B-cell epitopes from novel H1N1 are conserved in recent seasonal strains (and these may be inaccessible due to shielding by glycoconjugates as we discuss here), while the majority (69%) of CD8+ T-cell epitopes are conserved [30
]. Unpublished data from author JAM [Wanzeck K and McCullers JA, in preparation] demonstrate that the glycosylation status of the influenza virus hemagglutinin profoundly affects disease pathogenesis during sequential infection of mice. First infection with a highly glycosylated influenza virus leads to a sub-optimal antibody response that does not protect mice from re-infection with a poorly glycosylated variant of the same strain. Re-infected mice clear the challenge viruses but show severe morbidity and pathology following viral clearance attributable to aberrant CD8+ responses.
We propose that the relative paucity of HA glycosylation has two effects on the pathogenesis of novel H1N1. First, it has been demonstrated previously that lack of glycosylation contributes to virulence in the lung by precluding clearance by innate factors [25
]. This likely contributed to the extreme virulence of the 1918 pandemic strain, wherein a large part of the virulence in animal models has been attributed to the HA and NA [31
]. Similarly, novel H1N1 is highly virulent in animal models, a property linked to its ability to replicate efficiently in the lungs [32
], and likely attributable to lack of glycosylation. Second, we suggest that Original Antigenic exposure to H1N1 viruses with highly glycosylated HA1 could generate an immune response similar to that seen in mice exposed to highly glycosylated mutants. This immune response would not protect from infection, but would permit viral clearing and introduce the possibility of CD8+ cell-mediated toxicity in some subset of those affected. Therefore, persons whose first exposure to H1N1 viruses was to the highly glycosylated re-emergent viruses that appeared in circulation after 1977 would have a much lower level of immunoprotection, and would be at risk for aberrant T-cell responses and severe lung disease from cytokine storm. In support of this thesis, an autopsy series of 21 patients with fatal novel H1N1 infection demonstrated severe lung pathology highlighted by an aberrant pulmonary immune response characterized by a massive infiltration of CD8+ T-cells into the lungs [33
For this type of immuno-failure, a mismatch of antibody and T-cell response is required. Therefore, a gradation of responses would be expected within this age range. An individual exposed previously only to H1N1 viruses with highly glycosylated HA1 could have experienced the appropriate T-cell epitopes enough times to have a robust CD8+ response, but have little to no cross-protective antibody to novel H1N1. If such persons were exposed relatively infrequently and to few distinct variants of H1N1, the mismatch would be especially great. In our composite age distribution of cases, the peak occurs in the age range 10-19 years. A shift in case and age distributions to younger age appears to be a general characteristic of influenza pandemics [34
], and there is now a general consensus that school age children may be important early spreaders of pandemic disease. However, there is more detailed work [e.g., [35
]] that suggests that pre-school children should lead and possibly dominate case distributions. Therefore, while it may be that this extraordinary concentration derives primarily from the high social contact rate in this age group coupled with some proclivity of this particular influenza virus, we should also admit the possibility that additional immunological factors could be contributory. In the US between 1977 and 1988, H1N1 and influenza B viruses dominated in circulation in 8 of the 12 seasons. Between 1989 and 1998, H1N1 and B viruses dominated in only 3 of 10 seasons whereas from 1999 to 2008, H1N1 and B viruses dominated in 4 of 10 seasons and H1N1 viruses circulated widely but did not dominate in 2007/8 [36
]. There is also evidence that genetic divergence within H1N1 viruses was lower in the 1990s than in either the 1980s or the first decade of the 21st
]. Taken together, persons currently age 10-19 would have the smallest opportunity to encounter H1N1 viruses as their Original Antigenic Sin, and their exposure would have been to very similar viruses. Older individuals would have a greater opportunity for OAS exposure to more genetically diversified H1N1 viruses, though exposure for all of these would have been to a highly glycosylated HA. Young children, relatively naïve with respect to influenza, would have little opportunity to develop a robust T-cell response to H1N1, and no prospect for antibodies neutralizing to novel H1N1. They could be productively infected producing mild disease without the possibility of T-cell mediated immunopathology. A recent study, also in mice [38
], demonstrated that regulation of the cytotoxic T-cell response in highly pathogenic influenza infection may require a fine hand for effective modulation. Complete suppression of the cells required for proliferation of CD8+ T-cells in the infected lung was not effective in reducing morbidity or mortality; but modulating the facilitating cell population using a widely available and inexpensive human pharmaceutical did substantially mitigate the impact of infection.
In summary, persons aged ≤ 32 years are expected to have an OAS experience restricted to highly glycosylated H1N1 viruses and would thereby be expected to have inadequate immuno-protection for exposure to novel H1N1. Furthermore, the subgroup of age 10-19 years could have had an especially low rate of OAS exposure to any H1N1 virus. If the mouse data map onto humans, an OAS response conditioned by a highly glycosylated H1N1 virus could be responsible for the morbidity seen in a subgroup of younger persons. About one-half of cases with severe pneumonia in Mexico had no underlying medical condition; and in a detailed study of a subset of these cases, no evidence was found for bacterial pneumonia as a co-morbid condition [39
]. The evidence we cite suggests that such cases should be examined for cytotoxic T-cell activity; and considered for experimental therapy with likely modifiers of the influenza-specific inflammatory process, such as agonists of peroxisome proliferator-activated receptor-γ [38
The age group, 32< age < 61 years: Experience, with a tincture of time
In our composite of countries, the incidence of cases in these age groups declined curvilinearly to very low levels. Persons in a subgroup, aged 52-61 years, were born during the last decade of H1N1 dominance. The H1N1 viruses of 1947 to 1957 were more highly glycosylated than those of earlier years, but there was considerable heterogeneity and genetic divergence making this exposure somewhat rich. In particular, the sites for potential glycosylation varied more significantly during this period than after the re-emergence in 1977 (Figure ). It seems reasonable, therefore, that immunoprotection derived from OAS exposure in this subgroup would be considerable, if distinctly lower than in those over age 61. The associated experience is captured in the age group 50-59 (Figure ).
It is more difficult to argue that those aged 32 to 52 years should have an acquired immunoprotection different from those age <32 years, since none born between 1957 and 1977 would have had exposure to an H1N1 virus as an OAS experience. We note, however, that three of the eight genes in novel H1N1 (M, NP, NS) are cited as derived from the common H1N1 ancestral line and a fourth, PB1, is thought to have been reassorted from human H3N2 viruses and thence from an avian precursor [42
]. Persons aged 32-51, therefore, would have had greater cumulative immunological experience with variation in 4 of the gene segments currently found in novel H1N1. It has also been previously noted that infection with H3N2 viruses is somewhat protective against reinfection with H1N1 viruses and vice versa [43
] This less specific immunological experience coupled with a non-H1N1 OAS response appears to have produced a susceptibility to novel H1N1 intermediate between the two age groups with a direct OAS response to H1N1 viruses.