Variation in protection of four divergent avian influenza virus vaccine seed strains against eight clade 2.2.1 and 2.2.1.1. Egyptian H5N1 high pathogenicity variants in poultry

Background Highly pathogenic (HP) H5N1 avian influenza virus (AIV) was introduced to Egyptian poultry in 2006 and has since become enzootic. Vaccination has been utilized as a control tool combined with other control methods, but for a variety of reasons, the disease has not been eradicated. In 2007, an antigenically divergent hemagglutinin subclade, 2.2.1.1, emerged from the original clade 2.2.1 viruses. Objectives The objective was to evaluate four diverse AIV isolates for use as vaccines in chickens, including two commercial vaccines and two additional contemporary isolates, against challenge with numerous clade 2.2.1 and clade 2.2.1.1 H5N1 HPAIV Egyptian isolates to assess the variation in protection among different vaccine and challenge virus combinations. Methods Vaccination-challenge studies with four vaccines and up to eight challenge strains with each vaccine for a total of 25 vaccination-challenge groups were conducted with chickens. An additional eight groups served as sham-vaccinated controls. Mortality, mean death time, morbidity, virus, and pre-challenge antibodies were evaluated as metrics of protection. Hemagglutination inhibition data were used to visualize the antigenic relatedness of the isolates. Results and conclusions Although all but one vaccine-challenge virus combination significantly reduced shed and mortality as compared to sham vaccinates, there were differences in protection among the vaccines relative to one another based on challenge virus. This emphasizes the difficulty in vaccinating against diverse, evolving virus populations, and the importance of selecting optimal vaccine seed strains for successful HPAIV control.


Introduction
Use of vaccination for the control of avian influenza virus (AIV) has become routine, combined with other control measures such as improved biosecurity in areas where H5N1 highly pathogenic (HP) AIV is endemic in poultry. Historically, only a few H5 AIV seed strains have been used in commercial vaccines. Two of the most widely used have been A/goose/Guangdong/1/1996 which is the source of hemagglutinin (HA) and neuraminidase (NA) genes in the inactivated Re-1 vaccine 1 (use discontinued in China in mid-2008 2 although its use continued in Egypt and Vietnam) and A/Mexico/232/1994 H5N2 (Mex/94) low pathogenic (LP) AIV which is used in several inactivated commercial products. Currently, vaccines with the Mex/94 lineage virus and a reverse genetics produced vaccine with an Egyptian 2010 strain antigen are among the vaccines being used. Because of the ability of influenza A viruses to mutate and drift antigenically, there are reports of H5 field viruses which escaped immunity induced by these common vaccines in areas where they have been used widely, [3][4][5] including Egypt.
H5N1 HPAIV was first detected in poultry in Egypt in 2006. 6 As then vaccination has been one of the control measures utilized to help control the virus and reduce the impact of infection on poultry. However, H5N1 HPAIV outbreaks have continued to occur and vaccination alone has not been as effective as expected. One of the critical factors known to affect vaccine efficacy is antigenic matching between the challenge virus and vaccine. In countries that have used the same vaccines for an extended period of time, the field viruses have evolved to evade the immune response to the vaccine. 4,5,7 Improper vaccination 8 and inadequate coverage of poultry populations 9 make control of virus spread difficult which may contribute to the emergence of antigenic variants. In Egypt, a new variant, 2.2.1.1, (previously referred to as 'variant clade 2.2.1', 'sublineage B' or 'sublineage E') first described in 2007 is possibly an example of this. 6,10 The emergence of this clade could complicate vaccination strategies if the currently available vaccines have reduced efficacy against viruses in this clade.
Goals of this work include evaluating the efficacy of the current vaccines against numerous Egyptian isolates including isolates clade 2.2.1.1 and to determine if local, contemporary isolates from Egypt would provide any improvement in protection. Also, as experimental evaluations of AIV vaccines are often published later than is practical, a broader goal was to assess the consistency of protection against multiple challenge viruses, some of which were relatively closely related genetically and antigenically. To accomplish this, we have conducted vaccination-challenge studies with four vaccines and up to eight H5N1 HPAIV challenge viruses per vaccine.

Virus source and propagation
Virus isolates used for vaccine-challenge studies and antigenic cartography were obtained from either the repository at Southeast Poultry Research Laboratory, USDA-ARS (SEPRL; Athens, GA, USA), or the National Laboratory for Quality Control in Poultry Production (NLQP; Cairo, Egypt). All influenza isolates were propagated in specific pathogen-free (SPF) embryonating chicken eggs (ECE) by standard procedures. 11 Allantoic fluid containing infectious virus was used as challenge virus, and as the antigen source for vaccines and hemagglutination inhibition (HI) assays. The sequence data for the HA genes of all viruses used here have been published with analysis (isolates from 2006 to 2008) 6 or have been submitted to GenBank (isolates from 2009 to 2010).

Vaccines
Four viruses were selected for use as vaccine antigens. Two were selected because of their widespread use as commercial All vaccines were inactivated, whole virus, adjuvanted in an oil emulsion. The vaccine prepared with GS/GD/96 was the commercial product, Re-1 (Harbin Veterinary Research Institute, Harbin, China), 2 all others were experimental vaccines prepared in-house as described by Stone et al. 12 using 0Á1% beta-propriolactone inactivated allantoic fluid from ECE and adjusting the pH to 7Á0 with sodium bicarbonate. Sham vaccine was prepared with allantoic fluid from uninfected ECE. Viruses used to produce vaccines in-house were evaluated for HA titer by standard methods 11 and by titration in ECE as described previously. 11 The HA and 50% egg infectious dose (EID 50 ) titers for the viruses are given in Table 1.

Challenge viruses
Eight challenge viruses were selected from among H5N1 HPAIV viruses from poultry in Egypt, aiming to represent a diverse sampling of HA1 protein sequences.

Vaccination-challenge studies in chickens
Each vaccine-challenge group contained 10 individually tagged, SPF white leghorn chickens from SEPRL in-house flocks. Vaccine-challenge virus groups are listed in Table 2. Chickens were vaccinated between 3 and 4 weeks of age with 0Á5 ml of vaccine by the subcutaneous route. Three weeks after vaccination, the birds were bled for serum to evaluate antibody to both the vaccine and challenge virus by HI assay. The birds were also challenged 3 weeks post-vaccination with 10 6 EID 50 of HPAIV per bird administered in 0Á1 ml by the intrachoanal route. Oropharyngeal swabs were collected 2 days post-challenge (DPC) to evaluate virus shed titers by quantitative real-time RT-PCR (qrRT-PCR).

Quantitative real-time RT-PCR
Oropharyngeal swabs collected at 2 DPC were processed for qrRT-PCR to determine virus shed titers. RNA was extracted as described by Das et al. 13 using a combination of Trizol LS (Invitrogen, Inc., Carlsbad, CA, USA) and the MagMAX 96 AI/ND Viral RNA isolation kit (Ambion, Inc., Austin, TX, USA) with the KingFisher magnetic particle processor (Thermo Scientific, Waltham, MA, USA).
Quantitative rRT-PCR which targets the influenza M gene 14 was performed using the 7500 FAST Real-time PCR System (Applied Biosystems, Foster City, CA, USA) and the Hemagglutination inhibition assay Sera were tested by the HI assay to evaluate antibody levels to both the vaccine and challenge virus in vaccinated birds prechallenge. Because GS/GD/96, the source of the HA in the Re-1 vaccine was not available, a closely related isolate which was available was used: A/HongKong/156/1997 H5N1 (HK/ 156) (98Á1% amino acid identity in the HA1).
The assay antigens were prepared by inactivating infectious allantoic fluid with 0Á1% beta-propiolactone and adjusting the pH to 7Á0 with sodium bicarbonate. The HI assays were performed in accordance with standard procedures. 15 Titers were calculated as the reciprocal of the last HI positive serum dilution.

Antigenic cartography
A total of 23 H5N1 HPAIVs from Egypt were selected for characterization by antigenic cartography. Each virus used as a vaccine and/or challenge strain was included. Additional isolates were selected to represent divergent H5 HA1 protein sequences. In addition to the isolates from Egypt, five H5N1 isolates from different HA clades were included (the HK/156 antigen was used to approximate GS/GD/96). Two classical H5 clade viruses were also used, one of which was the Mex/94 isolate included as one of the vaccines in this study.
Reference sera were produced with an inactivated, oil emulsion vaccine prepared with the Montanide ISA50V adjuvant (Seppic, Inc., Fairfield, NJ, USA) using allantoic fluid prepared as described under HI assay. Three-week-old SPF white leghorn chickens from SEPRL in-house flocks were inoculated with 0Á5 ml of vaccine by the subcutaneous route, and serum was collected 3 weeks post-vaccination. Sera were treated with 5% chicken red blood cells for 30 minutes at ambient temperature to adsorb non-specific agglutinins. The HI assay was performed by standard procedures. 15 The antigenic map for the test antigens was constructed using matrix completion-multiple dimensional scaling (MC-MDS). 16 Given the HI table h with n antigens (n = 30 in this study) and m antisera (m = 28 in this study), each observed value h ij was transformed into h ij ¼ dH ij e À log2ðH i =H ij Þ, where H ij is the maximum HI titer in the HI table h and H j is the maximum HI titer for serum j, 1 ≤ i ≤ n, and 1 ≤ j ≤ m. The value of 2 was defined as a low reactor, and a value of 4096 was used for those entries with '>4096'. A low rank matrix completion algorithm was applied to identify those entries without experimental data and to reduce the noise in HI table. The antigenic distance was then calculated and mapped to a two dimensional antigenic map.

Statistical methods
Shed titers by vaccine and challenge virus, mean death times, and antibody titers to the vaccine and challenge virus prior to challenge were tested for statistical significance using the one-way ANOVA, all pairwise (Tukey test). If normality failed then the Student-Newman-Keuls method was used (pairwise). Linear regression was used to evaluate the correlation between antibody levels to the vaccine or challenge virus and virus shed titers. All statistical analyses were conducted with SIGMAPLOT 11.2 (Systat Software, Richmond, CA, USA). A P value of <0Á05 was considered to be significant. Statistical groups are denoted in tables as superscript and on graphs by lowercase letters. Groups with different letters are significantly different at a P value of <0Á05. Groups apply to values in the same column. Samples which were negative by rRT-PCR were counted as below the limit of detection (titer = 10 EID 50/ml).

Serology
Antibody levels to each vaccine 3 weeks post-vaccination (immediately prior to challenge) were evaluated by HI assay (Figure 1). The Re-1 vaccine (commercial product) and the in-house vaccine produced with the Mex/94 isolate produced significantly higher levels of antibody (geometric mean titers) than either of the experimental vaccines produced with the Egyptian H5N1 isolates: Egy/202/07 or Egy/13/08. Antibody levels using the challenge viruses as test antigen in the HI assay for each group varied widely by vaccine-challenge virus combination, but were consistently lower than antibody levels to the vaccine (in all non-homologous groups) ( Figure 1). Amount of virus shed was evaluated against antibody titer to either the vaccine or challenge virus immediately prior to challenge. There was no correlation between the level of virus shed and the antibody titer to either the vaccine or challenge virus for any of the treatment groups. There was also no association between antibody titers and mortality.

Vaccination challenge
Four vaccines were evaluated with up to eight H5N1 challenge viruses from Egypt for a total of 25 vaccinechallenge virus combinations and eight additional sham vaccinated groups (Table 2). Protection was measured as prevention of mortality and decrease in virus shed. There was 100% mortality in all sham vaccinated groups and mortality varied by vaccine-challenge virus combination (Table 2). There was not a clear trend for prevention of mortality in groups vaccinated with Egy/202/07 or Egy/13/08 based on   Oral virus shed at 2 DPC varied by vaccine and challenge virus combination ( Figure 1); all vaccines significantly reduced shed compared to sham vaccinated chickens except for the group vaccinated with Egy/202/07 and challenged with Egy/13/08 (note that the groups challenged with Egy/ 202/07, Egy/44/09 could not be statistically evaluated versus sham vaccinates, because there was only one or two sham vaccinates, respectively, which survived long enough to be sampled for shed). All vaccines reduced shed similarly for seven of the eight challenge viruses; that is, the titers of virus shed was not significantly different among vaccinated birds for any virus, regardless of vaccine (Figure 1).
When virus shed was compared between the two clades for each vaccine, the differences in mean titers were significantly lower with challenge viruses from clade 2.2.1 versus challenge viruses from clade 2.2.1.1 for Re-1, Mex/94, and Egy/13/08 ( Figure 2 1 (Figure 3). The antigenic distances within clade 2.2.1 is 1Á4609 AE 0Á7470 (mean AE standard deviation) antigenic units (an antigenic unit is equal to a twofold difference in HI assay titer), and within clade 2.2.1.1 is 1Á6291 AE 1Á3042 units. However, the intercluster distance is 2Á1875 AE 0Á9904 units. The antigenic distances within each clade (2.2.1 or 2.2.1.1) are statistically smaller than those between these two clades, P < 0Á001 (this result is consistent to those reported previously 17 ). The antigenic distances between the vaccine strains and challenge viruses are shown in Table 3.

Discussion
As expected, minor differences in protection (reduction of mortality and reduction of virus shed) based on vaccine and challenge virus combination were observed. Also there were higher mortality and shed titers in the Re-1 vaccinated groups that were challenged with clade 2.2.1.1 isolates versus clade 2.2.1 isolates. This data support the hypothesis that antigenic drift of H5N1 HPAIV in Egypt may be a result of immune pressure, and highlights the need for updated vaccines. In mid-2008, Re-1 was discontinued in China 2 where was replaced by a vaccine with a more recent H5N1HA gene, although, use of Re-1 continued for at least another year in Egypt. Continued use of Mex/94 based vaccines should also be critically reviewed; collective analysis of the  two clade 2.2.1.1 challenge isolates tested here showed significantly more virus shed versus the clade 2.2.1 challenge isolates, although mortality was not significantly different. The protection provided by Mex/94 will likely continue to deteriorate from its current levels similar to what has been noted with long term use of AIV vaccines in poultry in the past. [3][4][5] Our results are similar to what was reported by Grund et al. 18 based on protection by clade, although they used different challenge viruses from Egypt (one from each clade) and three of their four vaccines were different (Mex/94 was used commonly). Grund  Although there are differences among the vaccines in overall reduction of virus shed based on challenge virus clade, individually 17 of 18 vaccinated groups (the seven groups challenged with either Egy/202/07 or Egy/44/09 were not included in the statistical analysis because too few sham vaccinates survived long enough to be tested to determine the mean shed titer with adequate power) did reduce shed significantly versus sham vaccinates. Individually, 24 of 25 of the vaccine challenge groups also had significantly lower mortality when compared to sham vaccinates challenged with the same isolate. This has important implications for evaluating vaccines because most vaccines will provide measurable protection against mortality when compared to non-vaccinated or sham vaccinated birds in an experimental setting, but optimal protection may only be recognized in comparison to less protective vaccines. Furthermore, protection is always better in the lab than in the field; small differences in an experimental setting may be amplified in the field.
Somewhat interestingly, there was no significant difference in the number of chickens shedding detectable levels of virus based on vaccine or challenge virus clade. There was one challenge virus, Egy/20/09, for which significantly fewer chickens shed virus with three of the four vaccines (the exception was Mex/94). This virus also had the longest MDT with the shams, although this was only significant as compared to two other isolates. This suggests that vaccine efficacy may differ even among relatively closely related challenge viruses due to minor biological variations; if this isolate had been the only one used most of the vaccines would have appeared to have been more efficacious at reducing shed than they were. Because halting the transmission is a primary goal of vaccination, the optimal vaccines will substantially reduce both the shed titers and number of animals shedding virus. Interestingly, through 2012 the 2.2.1  19 reinforcing that the selection pressures on influenza are numerous (e.g., immune pressure, biological, epidemiological), and complex. The serological results were also interesting in that antibody titers to neither the vaccine nor the challenge virus were predictive of either survival or shed. This is in contrast to the somewhat loosely held principle that HI titers over 40 indicate that a bird will be protected from mortality 20 and greater than 120 prevent replication. 21 However, others have reported good protection against mortality in groups when geometric mean titers were below an HI titer of 40 [22][23][24] and it has been recognized that this cut-off is less predictive with vaccine viruses which are not well matched to the challenge virus among other factors. 25 This may be because HI assay is not measuring antibody to all protective epitopes, such as antibody to the fusion domain of the HA protein 26 or some other immunological component induced by inactivated adjuvanted vaccines.
Serology also revealed that Re-1 and Mex/94 induced higher antibody levels than either of the Egyptian isolates. Importantly, the differences in antibody levels did not translate into differences in protection. The lower immunogenicity of the Egyptian isolates is probably not due to vaccine formulation since the Mex/94 vaccine was prepared in-house identically to the Egyptian virus based vaccines and antigenic masses were similar based on HA titer. Therefore, there may be an inherent difference in the immunogenicity of the HA proteins of these isolates for chickens. Poor immunogenicity may be under selection to evade the immune response and may help explain why there was mortality with Egy/202/07 homologous vaccination challenge. Poorly immunogenic HA protein is thought to be why vaccines produced with A/chicken/Pakistan/NARC-1/ 1995 H7N3 produced poor immunity in the field and protected poorly against homologous challenge in experimental studies. 27,28 Conversely, good immunogenicity may explain why some isolates can provide excellent protection against relatively unrelated challenge viruses (e.g., Re-1 and Mex/94 vaccines with clade 2.2.1 challenge viruses). 29 As antigenic proximity did not correlate with protection based on the HI assay data, these data suggest that immunogenicity may be at least as critical as antigenic matching when selecting vaccine seed strains. Importantly, the observed segregation based on clade has been previously reported. 17 With this data HI based antigenic profiling alone is not sufficient to draw more general conclusions and nor is it clear what the effect of adjuvant may be on antibody reactivity. Finally, when updating vaccines, protection needs to be evaluated carefully, as simply using a recent local virus of the same genetic lineage may not always provide optimal protection.