Introduction

Influenza viruses belong to family Orthomyxoviridae, and they are divided into influenza A, B, and C types [8]. Influenza A virus contains eight genes that code for structural or non-structural components, including the surface glycoproteins, hemagglutinin (HA), and neuraminidase (NA), both of which are important components that allow the virus to adhere to host-cell receptors. Influenza A virus is thought to infect a variety of species of mammals and birds, including humans, pigs, waterfowl, and poultry [29].

Until recently, humans were infected by only a limited number of influenza viruses subtypes, such as H1N1 [21], H1N2 [6], H2N2 [10], H3N2 [4], H5N1 [25], H7N3 [28], H7N7 [7], and H9N2 [20]. Infection with H5N1 influenza A virus occurred in humans in 1997 in Hong Kong, causing deaths and leading to severe problems [25]. The infection was considered as a rare event; however, H5N1 influenza infection in humans is currently on the increase [32], suggesting that an influenza pandemic caused by H5N1 or another subtype will occur in the human population someday [30]. On the other hand, the H5N2 subtype of influenza virus has been shown to be virulent in poultry, but humans do not seem to be susceptible to it [1]. In a previous study, no detectable anti-H5 antibody was found in human sera, and no viruses were isolated from humans when an H5N2 influenza outbreak occurred in poultry in Italy [5]. In 2005, the same subtype of virus, A/H5N2, was isolated from chickens raised on a farm in Ibaraki prefecture, Japan [14]. The genetic properties of these viruses were very similar to those of strains obtained from Latin America [9]. This outbreak resulted in the culling of almost six million chickens, and it took almost 1 year to suppress this outbreak [15].

In previous experiments, the results of serological tests have suggested that some employees engaged in epidemic control might have anti-H5 antibodies. It is expected that these employees were exposed to or infected by the H5N2 influenza virus; this was the first case that suggested the possibility of infection by H5N2 virus in humans [13, 17]. Unlike the case of H5N1 infection, a “routine” protocol for serological diagnosis of H5N2 virus infection in humans has not yet been developed, suggesting that an established method for detecting anti-H5N2 antibody in humans is required. One possible candidate method for detecting anti-H5 antibody is a microneutralization (MN) test (MNT), because of its high specificity and sensitivity for detecting H5N1 virus infection in humans [19]. However, an examination conducted previously revealed positive results in an MNT, i.e., viral agents were not isolated from these employees; therefore, unknown agents, including the characteristics of the patients or other unknown agents, may influence the MN titer by unknown mechanisms. In this study, MNT and hemagglutination inhibition (HI) tests were performed, and the results were compared for three human populations: (1) general inhabitants of the geographical area where the H5N2 outbreak occurred, (2) general inhabitants of Japan, and (3) employees who worked in or were engaged with the poultry industry or had jobs related to it. The serological data on these three populations were analyzed to investigate the relationship between several factors.

Materials and methods

Sera

A total of 114 serum specimens were collected from inhabitants living in Ibaraki prefecture from May to August 2006. Through a questionnaire, it was verified that these inhabitants were not in direct contact with birds, including poultry. To represent the Japanese healthy population, 100 human serum samples were provided from the Serum Bank of the National Institute of Infectious Diseases (Tokyo, Japan). They were taken from healthy Japanese people living outside of Ibaraki prefecture during 2005. In addition, 52 serum samples were collected from employees working at H5-influenza-free poultry farms located in Ibaraki prefecture during August 2006. Their age distribution and genders are shown in Table 1. Before serological examination, sera were treated with receptor-destroying enzyme (RDE II; Denka seiken, Tokyo, Japan) to remove factors that cause non-specific reactions.

Table 1 Properties of human populations analyzed in this study
Full size table

Cells and viruses

Madin–Darby canine kidney (MDCK) cells were used for MNT. Cells were grown and maintained in Eagle’s minimum essential medium (E-MEM, Invitrogen, California, USA), supplemented with 10% fetal bovine serum (Invitrogen), 2 mM l-glutamine (Invitrogen), 100 U/ml penicillin–streptomycin (Invitrogen), and 0.25 μg/ml fungizone (Invitrogen). For MNT, viruses were grown in E-MEM supplemented with 2 mM l-glutamine, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES; Invitrogen), 1% bovine albumin (SIGMA, St Louis, USA), 100 U/ml penicillin–streptomycin, and 10 μg/ml acetylated trypsin (SIGMA). The influenza virus strain A/chicken/Ibaraki/1/2005 (H5N2) was provided by the National Agriculture and Food Research Organization, National Institute of Animal Health (Ibaraki, Japan). The virus was propagated in 10-day-old hen eggs.

MNT

MNT was carried out according to methods described elsewhere [31]. Treated sera were mixed with equal volumes of virus antigens [prepared at 1.0 × 102, 50% tissue culture infection dose (TCID50)/50 μl], and incubated at 37°C for 30 min. The mixture was transferred onto MDCK cells prepared on 96-well flat-bottom plates (Greiner GmBH, Germany) and incubated at 37°C and 5% CO2 for 96 h. After incubation, infected cells were fixed with PBS(−)-10% formalin and stained with naphthol black (Sigma) solution. The optical density was determined by measuring the absorbance at 630 nm using a microplate reader (Model No. 680; Bio-Rad Laboratories, California, USA). Neutralization titer was expressed at the highest dilution of serum that protected TCID50. Examinations were repeated at least twice.

HI test

The HI test was performed according to methods described previously [31] with 0.75% horse erythrocytes in phosphate-buffered saline [PBS(−)] supplemented with 0.01% gelatin, 0.5% chicken erythrocytes in PBS(−), and 0.5% turkey erythrocytes in PBS(−). The erythrocytes were purchased from Nippon Bio-test Laboratories Inc. (Tokyo, Japan). HI titer was expressed at the highest dilution of serum that showed hemagglutination inhibition. Examinations were repeated at least twice.

Statistical analysis

MN and HI titers below the lower limit of these assays were given an arbitrary intermediate value of five [24]. Differences between each population and differences in gender were compared by Fischer’s exact probability test, and differences in age were tested by Mann–Whitney’s U test. Probability (P) values of less than 0.05 were considered to indicate statistical significance.

Informed consent and human experimental guidelines

Before this study was implemented, a research committee was established in the Ibaraki Prefectural Government, and informed consent was obtained from each subject.

Results

MNT

To investigate the distribution of anti-H5N2 influenza neutralizing antibody, MNT was performed using human sera collected for this study. Among Ibaraki inhabitants (II, n = 114), eight samples showed an MN titer of ≥40 (male = 6, female = 2). Among the healthy Japanese population, the national average (NA, n = 100), four samples showed an MN titer of ≥40 (male = 1, female = 3). Further, among poultry workers (PW, n = 52), eight samples showed anMN titer of ≥40 (male = 5, female = 3). The highest titer, 160, was observed in two samples (Table 2).

Table 2 Results of microneutralization test
Full size table

HI test

HI tests were performed to investigate the existence and distribution of anti-H5 hemagglutination inhibiting antibody using horse erythrocytes. Among II, nine samples showed an HI titer of ≥40 (male = 6, female = 3). Among NA, two samples showed an HI titer of ≥40 (male = 1, female = 1), and among PW, two samples showed a titer of ≥40 (male = 1, female = 1). In this examination, the highest titer, 80, was observed in two individuals (Table 3).

Table 3 Results of hemagglutination inhibition (HI) test with horse erythrocytes
Full size table

Statistical analysis of serological data

Table 4 describes the geometric mean titers (GMTs) of the MNT and HI tests of human sera samples. In II, GMTs of the MNT and HI tests were 12.45 and 16.46, respectively. In PW, GMTs were 11.58 in MNT and 8.99 in the HI test. In NA, GMTs were 7.90 in MNT and 10.21 in the HI test.

Table 4 Geometric mean titers (GMTs) of microneutralization test and HI test with human sera
Full size table

A recent investigation demonstrated that the cutoff value for the anti-H5 influenza MN titer to be judged as positive was ≥40 [18, 27]. Based on these findings, MNT titer of ≥40 was designated as positive. In this study, positive MNT results were found in 7.0% in IP, 4.0% in NA, and 15.3% in PW. The data were analyzed to investigate the relationship between seropositivity in the MNT and variable factors, as shown in Table 5. In PW, seropositivity was related to age (≥50 years old) with statistical significance (P = 0.038). Gender was not related to seropositivity. In II and NA, seropositive results were not related to age or gender.

Table 5 Statistical analysis of MN titer
Full size table

Serological assays of specimens showed an MN titer of ≥40

Twenty specimens (eight from II, four from NA, and eight from PW) showed an MN titer of ≥40; these specimens were designated as seropositive in this study. Table 6 describes the result of Chi-square analysis of II and NA, and PW and NA. The number of positives in PW in comparison to that in NA was statistically significant (P = 0.018). Table 7 shows the comparison of MN and HI tests for each positive specimen. Among these specimens, the HI titer with horse erythrocytes ranged between <10 and 40. Specimens II-18 and II-83 showed a titer of 40 with turkey and chicken erythrocytes; however, the remaining specimens showed a titer of <10.

Table 6 The number of seropositive results based on the results of microneutralization test in three populations
Full size table
Table 7 Serological assay responses of the individuals that showed a microneutralization titer of ≥40
Full size table

Discussion

In an H5N2 virus outbreak among poultry farm chickens in Ibaraki, Japan, the results of MNT suggested that some people who engaged in epidemic control had anti-H5 antibody [27]. This was the first case that suggested H5N2 virus infection in humans [13, 17]. It resulted in the culling of about six million chickens [17] and raised serious concerns with regard to human health.

In this study, it was shown that HI titers of human samples in assays with horse erythrocytes, were higher than those of avian species. When HI titers with horse and avian species were compared (Table 7), the difference between the titers with horse erythrocytes and those with avian erythrocytes was statistically significant in PW (P = 0.0156 with respect to turkey and chicken erythrocytes, respectively, as determined by Wilcoxon signed rank test). However, the difference was not significant in II, possibly due to the presence of HI titers higher than those with horse erythrocytes. Interestingly, II-18 and II-83 showed a titer of 40 in HI tests with turkey and chicken erythrocytes; these values were higher than those obtained with horse erythrocytes. The reason why these specimens yielded higher titers is not clear. One possible explanation is that RDE treatment of serum samples was insufficient; however, this explanation requires further examination using techniques such as western blotting or H5-specific ELISA with purified antigens.

In a previous study, Rowe et al. [19] suggested that the MN test was more sensitive and specific than HI tests when H5N1 influenza viruses were used as antigens, and Stephenson et al. [23] showed that, in an HI test, titers with horse erythrocytes were higher than those with turkey erythrocytes when H5N3 subtype was used as antigen. In this study, it was shown that the MN titer was higher than that of the HI test when H5N2 virus was used as antigen and that the HI titer with horse erythrocytes was higher than that with avian erythrocytes. This is the first report that suggests the occurrence of the phenomenon described above with H5N2 influenza virus. A previous study showed that detectable antibody responses were not observed when other subtypes of influenza viruses such as H4, H6, and H10 were inoculated to human volunteers, but viruses were recovered from them [2]. The results presented in this study also suggest that the occurrence of these phenomena may be related to specific interactions between H5 viruses and immune responses in humans; however, further study is required to confirm this. In Table 4, it is shown that the GMTs of the MNT and HI data obtained from II (P < 0.0001 in MNT and P < 0.05 in the HI test, respectively), in MNT from PW (P < 0.05) were significantly higher than that of NA. These results suggest that inhabitants and poultry workers living in Ibaraki prefecture may possess higher levels of neutralizing antibody or hemagglutination inhibiting antibody than the Japanese healthy population. Furthermore, these results also showed that 20 specimens had an MN titer of ≥40 (Tables 6, 7) and also indicated that the number of positive serum neutralization antibody titers against avian influenza virus was significantly higher (P = 0.018) in PW than in NA (Table 6). These results suggest that PW may possess a higher MN titer and might be exposed to the H5 subtype of influenza virus. It has been epidemiologically suggested that individuals working in specific occupations, such as swine farmers, poultry workers, and veterinarians, are exposed to some subtype of influenza virus [3, 12, 16]. It has also been suggested that people living in southeastern China might be affected by or exposed to several subtypes of influenza virus [22]. These results suggest that the specific occupations described above or geographic factors may influence the exposure to some subtypes of influenza virus. In Ibaraki prefecture, it has been estimated that about 11.9 million laying chickens are raised [26]. Of these, about two million are chickens and ten million are hens. This value was the greatest in Japan in 2004. This would suggest the possibility that inhabitants living in this region may come into contact with or be exposed to chickens or other fowl more frequently. There is also the possibility for inhabitants being exposed to chickens or other fowl being higher than that for others living in other regions in Japan. However, the increase in antibody titer should be examined with other examinations to determine whether the results are “positive” or “false-positive”. This suggests that further study will be required to demonstrate the presence of anti-H5 antibodies in human specimens.

Our data show that age (≥50 years old) was related to seropositivity with statistical significance (P = 0.038, Table 5). However, in previous studies, age was not related to elevated antibody titers in the HI test [11] or in MNT [12]. These findings suggest that our data may be influenced by an insufficient number of specimens or by other unknown factors. It is suggested that further study will be needed to confirm these results—for instance, by increasing the number of specimens.

The results shown in this study suggest that antibodies specific for H5 influenza viruses were present in non-exposed human sera. A previous study using an MN test and single radial haemolysis (SRH) showed that anti-H5 antibodies were not present in human sera. The failure of isolation of viruses from humans showed that avian-to-human transmission of virus did not occur [5]. However, it is believed that avian-derived viruses may be able to replicate in humans when they are inoculated at high doses [2], suggesting that the invasion of H5N2 viruses into humans could occur: this also suggests that further laboratory-based surveillance is required. This would enable the detection of “infection” or “exposure” in humans with viruses unrelated to human strains, including avian influenza viruses. For further study, the evaluation of serological results should be confirmed by other assays, i.e., western blotting and specific ELISA using purified antigens. It is strongly suggested that isolation of H5N2 viruses from humans must also be carried out, which would be sufficient for confirming infection in humans.