Cross-subtype Immunity against Avian Influenza in Persons Recently Vaccinated for Influenza

Seasonal influenza vaccination may induce heterosubtypic immunity against avian influenza virus (H5N1).

I nfl uenza viruses are segmented, negative-sense RNA viruses belonging to the family Orthomyxoviridae. According to the antigenic differences in nucleoprotein and matrix proteins, 3 types of infl uenza viruses (A, B, and C) have been described. Infl uenza viruses A and B are associated with seasonal illness and death, whereas infl uenza virus C causes mild infections (1,2). Infl uenza A viruses are subtyped on the basis of the antigenic differences on external hemagglutinin (HA) and neuraminidase (NA) glycoproteins. Human type A infl uenza virus subtypes have been limited to H1, H2, and H3 and to N1 and N2 (3). Several HAs and NAs have been isolated from avian hosts; occasionally, they have been associated with human outbreaks (4,5).
Cytotoxic T lymphocytes play a central role in the clearance of primary infl uenza virus infection, peaking after 7-10 days; the peak in antibody titers occurs 4-7 weeks after primary infection (6)(7)(8). Neutralizing antibodies are completely protective against secondary challenges only with closely related strains, but they are ineffective against viruses with major antigenic divergence. For this reason, current infl uenza vaccines are prepared annually on the basis of World Health Organization forecasts on the most probable infl uenza A and B virus strains thought to be circulating in the next seasonal outbreak (5,7). By contrast, cellular responses to cross-reactive epitopes may provide a substantial degree of protection against serologically distinct viruses (9). The ability of infl uenza viruses to mutate and reassort their HA-NA genome segments between different animal species is a main concern because immunity generated by previous infections or vaccinations is unable to prevent infection by itself, although it may reduce virus replication and spread (8)(9)(10).
To date, 3 infl uenza subtypes have produced pandemic disease in humans: H1N1 in 1918, H2N2 in 1957, and H3N2 in 1968 (4,11,12). In 1997, during the avian infl uenza (H5N1) outbreak in Hong Kong Special Administrative Region, People's Republic of China, a cross-reactive cellular immune response induced by infl uenza (H9N2) was able to protect chickens from infl uenza (H5N1) (13). Moreover, adults living in the United States who were never exposed to H5N1 subtype have shown cross-type cellular immunity to infl uenza A virus strains derived from swine and avian species (including the H5N1 subtype isolated in Hong Kong) (14). Thus, speculation that cross-reactive T cells may decrease illness and death by reducing the replication of the new infl uenza virus, even if elicited by a different strain, is reasonable.
Avian infl uenza A viruses of the H5N1 subtype are currently causing widespread infections in bird populations. Numerous instances of transmission to humans have been recently reported in Asia and Africa, with the infection resulting in severe disease or death (>50% fatality rate). Hence, the aim of the present study was to evaluate the immune cross-reactivity between human and avian infl uenza (H5N1) strains in healthy donors recently vaccinated for seasonal infl uenza A (H1N1/H3N2). Our data indicate that infl uenza vaccination may boost cross-subtype immunity against infl uenza (H5N1), involving cellular or humoral responses or both.

Study Population
Healthcare workers wishing to receive seasonal infl uenza vaccination at the Spallanzani Institute (n = 42) were enrolled. The study was approved by the local Ethical Committee; all participants gave written informed consent. Baseline characteristics of the study population are reported in the Table. Blood samples were obtained before (t0) and 30 days after vaccination (t1). The vaccine formulation was Fluarix, an inactivated and purifi ed split infl uenza vaccine (GlaxoSmithKline, Verona, Italy). The antigen composition and strains were A/California/7/2004-H3N2; A/New Caledonia/20/99-H1N1; and B/Shanghai/361/2002. Each 0.5-mL vaccine dose contains 15 μg HA of each strain in phosphate-buffered saline and excipients. Vaccine was administered intramuscularly.

Cells, Viruses, and Antigens
Madin-Darby-canine kidney (MDCK) cells were maintained in Dulbecco modifi ed Eagle medium (DMEM) containing 10% fetal calf serum (FCS), and 2 mmol/L Lglutamine, at 37°C in a 5% CO 2 humidifi ed atmosphere. The infl uenza (H5N1) virus used was strain A/Hong Kong/156/97 (kindly provided by Paul Chan) (15). The virus stock used as challenge antigen in the hemagglutination inhibition (HI) assay was propagated in the allantoic cavities of 10-day-old embryonated hen eggs. The allantoic fl uid was harvested 48 h postinoculation and clarifi ed by centrifugation. Virus concentration was determined by HA titration as previously described (16), and the virus was stored at -80°C until used. The virus stock used in the microneutralization (NT) and in the cell-mediated immunity assays was propagated in MDCK cells, and the culture supernatants were collected 48 h postinoculation. The 50% tissue culture infectious dose (TCID 50 ), determined by titration in MDCK cells, was calculated by the Reed and Muench method (17).
In addition, the frequency of IFN-γ-producing CD4 T lymphocytes from each donor in the absence of any stimulation was used to calculate the background for each stimulation. The resulting background levels were very low in every experiment, and no differences were observed between samples obtained before (t0 0.03% ± 0.04%) and after vaccination (t1 0.01% ± 0.03%). The frequency of antigen-specifi c CD4 T cells for each study participant was calculated by subtracting the relative background levels at t0 and t1.
Cell-mediated immunity was considered positive when the net increase was >0.2%. Although retesting samples on separate occasions gave reproducible results, t0 and t1 samples for each participant were tested simultaneously to further reduce test variability.
Multiparametric fl ow cytometry was performed by using a FACSCanto fl ow cytometer (Becton Dickinson). A total of 300,000 live events were acquired, gated on small viable lymphocytes, and analyzed with FACSDiva software (Becton Dickinson). The instrument was routinely calibrated according to the manufacturer's instructions.

Microneutralization and HI Assay
The NT was performed according to a previously described procedure (20), in agreement with indications from the World Health Organization (21) and the US Department of Health and Human Services (22). Specifi cally, 2fold serial dilutions of heat-inactivated (30 min at 56°C) human sera were performed in 50 μL DMEM without FCS in 96-well microplates. An equal volume of infl uenza virus (H5N1) (10 3 TCID 50 /mL) was then added to each well. Uninfected-cell wells, incubated with each test serum, were included in each plate as negative controls. After 1 h incubation at 37°C, the mixtures were transferred on MDCK cell monolayers and adsorbed at 37°C for 1 h. After washing, DMEM was added, and the plates were incubated for 2 days at 37°C in 5% CO 2 . NT titer was assessed as the highest serum dilution in which no cytopathic effect was observed by light microscope inspection. All serum specimens were tested in duplicate, and t0, and t1 samples from each patient were assayed in the same plate at the same time. The results were scored by persons blinded to the study participant's identifi cation. The test results were reproducible because random replication of the assays on independent occasions gave consistent results.
The antibody titer was also established by HI test, using for challenge either the seasonal vaccine or the egg-derived infl uenza (H5N1) preparation. HI assays were performed in V-bottom 96-well plates with 0.5% chicken erythrocytes, as described (16).

Biosafety Laboratory Facilities
All experiments with live highly pathogenic avian infl uenza A virus (H5N1) were conducted by using Biosafety Level 3-plus (BSL3+) containment procedures (23). All investigators were required to wear appropriate masks with HEPA fi lters.

Cell-mediated Immunity to Infl uenza Viruses
The frequency of circulating antigen-specifi c CD4 T cells in healthy donors enrolled in the study was analyzed by fl ow cytometry, by using intracellular cytokine staining assay after the in vitro expansion of effector cells. To generate effector cells from their memory precursors, PBMC were challenged with antigen in vitro for 3 days and expanded for 6 additional days in the presence of IL-2 (18).
Effector cells were characterized for their ability to release IFN-γ when cultured overnight in the presence of antigen. CD4 T cells were gated and analyzed for IFN-γ and IL-2 cytokine expression. A representative experiment with PBMC from a recently vaccinated healthy donor is shown in Figure 1. Without stimuli, no cytokine production in CD4 T cells was detected (Figure 1, panel A). However, the stimulation with the seasonal infl uenza vaccine preparation induced the production of IFN-γ by CD4 ef-fector T cells (Figure 1, panel B: 3.2% of IFN-γ+ CD4+ T cells). Stimulation with inactivated infl uenza (H5N1) virus induced a CD4 T-cell response (Figure 1, panel C: 1.0% of IFN-γ+ CD4+ T cells). Finally, some CD4 T cells specifi c for a pool of H5 and N1 (H5/N1) peptides were also generated in this donor (Figure 1, panel D: 0.6% of IFN-γ+ CD4+ T cells). No IL-2 production was observed in these experimental conditions.

Increased Cell-mediated Immunity after Seasonal Infl uenza Vaccination
When the extent of CD4 T-cell-mediated immunity before and after seasonal infl uenza vaccination was compared in the healthy donors enrolled in the study, a nonhomogeneous pattern of responses was detected (online Appendix Figure; available from www.cdc.gov/EID/content/14/1/121-appG.htm. After vaccination (t1), a 2-fold variation of the frequency of antigen-specifi c T cells higher than baseline was arbitrarily considered signifi cant. According to this threshold, an increased frequency of IFNγ-producing CD4 T cells specifi c for vaccine preparation was observed after vaccination in 5 (donors 8, 11, 17, 26, 42) of 21 donors (23.8%). A slight increase of frequency of the vaccine preparation-specifi c CD4 T cells was observed in 5 donors (donors 9, 12, 33, 36, 40; 23.8%); a mild-tosignifi cant decrease was observed in the remaining donors (n = 11; 52.3%).

H5 versus N1 Specifi city of the Cell-mediated Response
Because some study participants were reactive to inactivated infl uenza virus (H5N1) as well as to a peptide pool composed of 2 peptides from H5 and 2 from N1 consensus sequences, we analyzed whether this reactivity was preferentially directed against HA or NA. As shown by PBMC from a representative donor in Figure 2, the frequency of IFN-γ-producing CD4 effector T cells was appreciable after challenge with the inactivated infl uenza virus (H5N1) (Figure 2,  No specifi c CD4 T cells producing interferon-gamma (IFN-γ) were observed after challenge with H5 peptides (D). As negative control, either mock-infected culture supernatants or irrelevant peptides were used, giving results very similar to unstimulated cultures (not shown). A similar pattern was observed in 4 other study participants, supporting the hypothesis that the actual target of cross-subtype immunity could be N1.
pattern was observed in 4 other study participants, which supports the hypothesis that the target of cross-subtype immunity could actually be N1.

Increased Humoral Immunity after Seasonal Infl uenza Vaccination
Human sera from the same donors were tested for HI activity against both vaccine and infl uenza (H5N1) preparations and for neutralization activity against infl uenza (H5N1) virus. Individual titers are reported in Figure 3. A 4-fold rise in HA antibody titer is considered noteworthy, and after vaccination most donors (28/38; 73.7%) showed a noteworthy rise of HI titers against vaccine preparation, as indicated by an asterisk (Figure 3, top panel, black bars). HI titers against infl uenza virus (H5N1) remained at undetectable levels after seasonal vaccination (data not shown), but a rise of neutralization titer >20-fold over baseline was observed in 13 (34.2%) of 38 donors ( Figure 3, bottom panel, asterisk). All but 1 study participant also responded to seasonal vaccination by a rise in HI titers against vaccine preparation. One donor (21) showed high titers against the H5N1 subtype in NT but a low HI titer against vaccine, a unique situation in the study population. However, antibodies to both anti-infl uenza (H5N1) and infl uenza vaccine are raised by vaccination. Our fi ndings indicate that seasonal vaccination can raise neutralizing immunity against infl uenza (H5N1), which shows the existence of an antibodydependent cross-type immunity. No correlation between infl uenza-specifi c CD4 T cells and humoral responses was observed, which suggests that this type of antibody response was mainly CD4 T-cell independent.

Discussion
We observed that infl uenza-specifi c CD4-effector T cells could be generated by long-term cultures in vitro and easily monitored by fl ow cytometry as IFN-γ-producing cells. When this approach was used, a small frequency of CD4 T cells specifi c for H5N1 subtype could be detected in several persons at baseline. Seasonal vaccine administration may enhance the frequency of reactive CD4 T cells, boosting the cross-subtype cellular immunity against avian infl uenza (H5N1). We also observed that seasonal vaccination raised neutralizing immunity against H5N1 subtype in a large number of donors, showing the existence of an antibody-dependent cross-type immunity. Thus, cross-reactive immunity may involve cellular and/or humoral responses, but the humoral response seems to be CD4 independent.
From the present data, N1 appears to be 1 target for cross-type cellular immunity, although we could not rule out the involvement of different (i.e., internal) antigens as possible targets of immune recognition by effector CD4 T cells. Nevertheless, in animal models, cellular immunity (mainly CLT) targeting internal proteins (i.e., NP), partly responsible for heterosubtypic protection, was not induced effi ciently by inactivated vaccines (24). We did not use live virus, only inactivated split vaccine, whole inactivated virus, or HA and NA peptides for the infl uenza (H5N1) A/ Hong-Kong/156/97 strain. From our data, discriminating between the CD4 T-cell response against external or internal antigens in the case of vaccine preparation was not possible. For H5N1 subtype response, we can presume that the response is against the external antigens and that the results against peptides point to a specifi c response against NA.
Results obtained with the whole virus and those obtained with the H5 and N1 peptides are not in complete agreement (online Appendix Figure). This fi nding can be explained on the basis of the substantial differences in the antigen presentation underlying the whole virus and peptides. Moreover, we observed that a high activation of specifi c cells at baseline (t0) was associated with a reduced specifi c response after vaccination (t1), which suggests that stimulation of pre-activated T cells with high dose of antigen could induce T-cell anergy (25) with consequent loss of immune response. Preliminary evidence also suggests that humoral crosstype immunity is targeting antigens differently from HA: sera from persons showing signifi cant neutralizing titers against infl uenza (H5N1) did not recognize insect cells expressing HA from the H5N1 subtype (not shown) and did not show HI activity against H5N1 subtype. N1 may possibly also be a target of humoral immunity, but additional experiments such as Western blot analysis or inhibition of NA activity (26) are needed to clarify this point.
In animals, exposure to 1 specifi c subtype of infl uenza A virus can also induce protective immunity against challenges with other subtypes. This heterosubtypic or crossprotective immunity could represent a key mechanism for facing, and limiting, new infl uenza outbreaks. In 1997, during the Hong Kong infl uenza (H5N1) outbreak, an immune response induced by an infl uenza virus (H9N2), being T cells but not antibodies, protected chickens from lethal infl uenza (H5N1) (13). Moreover, adults living in an urban area of the United States have been described as having infl uenza-specifi c memory T cells that recognize epitopes of infl uenza A virus strains derived from swine and avian species, including the infl uenza (H5N1) strain involved in the Hong Kong outbreak in humans (14).
Our data confi rm that persons who have never been exposed to H5N1 subtype may be able to generate a cell-mediated response against the Hong Kong infl uenza (H5N1) isolate. This cross-type response may be naturally occurring (probably as a consequence of exposure to seasonal infl uenza strains).
In mice, both CD4 T-cell-independent and -dependent antibody responses contribute to the control of infl uenza virus infection (27,28). Although antibodies appear to facilitate the recovery from infl uenza infection, it is generally believed that B cells cannot produce neutralizing, isotypeswitched, infl uenza-specifi c antibodies in the absence of CD4 T-cell help (29,30). However, other data clearly demonstrate that B cells can also produce anti-infl uenza IgA, IgM, and IgG responses independent of CD4 helper T cells (27,31). A non-antigen-specifi c bystander response driven by activated CD4 T cells specifi c for heterologous antigen may contribute to so-called heterosubtypic immunity (8)(9)(10)12). However, the ability of infl uenza virus infection to promote B-cell activation and differentiation into shortlived, isotype-switched, antibody-secreting cells may result from a combination of B-cell receptor hypercross-linking, engagement of toll-like receptors, production of cytokines, as well as triggering of innate immunity.
In our study, cellular and humoral cross-reactive immunity seemed to target antigens other than HA. Infl uenza (H5N1) cases occur mainly in young people (32). This fi nding may be explained by hypothesizing that older people, although not previously exposed to H5N1 subtype, may have gained protective immunity by previous infections sustained by circulating infl uenza virus strains. It has also been shown that immunity to the N1 NA from the human infl uenza virus cross-reacts with the avian N1 NA virus and that this cross-reactivity protects mice against infection with the avian infl uenza virus (H5N1) (26). All these fi ndings may be explained by hypothesizing that cross-reactive immunity is targeting the N1 NA antigen. However, whether cross-reactive antibodies to NA and CD4 T cells would be protective against illness and death, especially from infl uenza (H5N1) infection is not known. Further studies will be necessary to elucidate this point.
In conclusion, we demonstrated that vaccination against seasonal infl uenza may boost a cross-reactive immunity against an unrelated strain responsible for deadly infections in humans, i.e., the avian infl uenza (H5N1) strain A/Hong Kong/156/97. These data, together with previous experimental results from mice studies and epidemiologic reports, indicate that cross-type immunity should be considered an important component of the immune response against novel infl uenza A infections.