New Methodological Approaches in The Development of Russian Live Attenuated Vaccine for Pandemic Influenza

Avian influenza viruses remain a major pandemic threat. In response to this threat, a number of pandemic vaccines have been developed. The objective of this paper is to review and summarize data from preclinical and clinical evaluation of Russian live attenuated influenza vaccines (LAIVs) against pandemic influenza based on cold-adapted A/Leningrad/134/17/57 (H2N2) master donor virus (MDV). The described LAIVs consist of reassortant viruses of 6:2 and 7:1 genomic composition (6 MDV genes: 2 WT genes and 7 MDV genes: 1 WT gene, respectively). Despite the differences in their genomic composition (6:2 or 7:1), LAIV candidates of H5, H7 and H2 subtypes acquired temperature sensitivity, cold-adaptation, and attenuation for different animal models. In addition, they were safe and immunogenic for healthy adult volunteers. The collected data indicate that 7:1 reassortants carrying HA genes of potentially pandemic viruses and the remaining genes from the MDV might be preferable pandemic LAIV candidates.


Introduction
Vaccination is considered an essential public health tool to control both seasonal epidemic and pandemic influenza. Several types of influenza vaccines (inactivated and live attenuated) and production technologies (egg-or cell derived vaccines and vaccines generated by reverse genetics) exist. World Health Organization (WHO) acknowledges benefits of vaccination with live attenuated influenza vaccine (LAIV) in controlling epidemic and pandemic influenza [1,2]. In contrast to inactivated vaccines, LAIVs are capable of inducing broad-spectrum and long-lasting immune responses, making them an attractive option for pandemic preparedness, especially in countries with very high population density. Furthermore, WHO considers the expansion of LAIV production as a promising strategy to increase influenza vaccine supply in pandemic situation [3].
Two live-attenuated, cold-adapted vaccines are currently manufactured in Russia (Microgen) [4] and in the USA (MedImmune) [5]. The LAIV consists of reassortant viruses, which contain hemagglutinin (HA) and neuraminidase (NA) gene segments from circulating wild-type (WT) viruses of interest on a backbone of the remaining six internal protein genes derived from the attenuated master donor viruses (MDVs) (genomic composition 6 MDV genes: 2 WT genes). A/Leningrad/134/17/57 (H2N2) (Len-MDV) and B/USSR/60/69 MDVs are currently used in Russia as MDVs for LAIV [6]. Although LAIVs have been under Reassortants for Russian LAIV are being produced by the Institute of Experimental Medicine (IEM, St Petersburg, Russia). IEM supplies the LAIV reassortants to the Russian manufacturer Microgen (Moscow). Besides that, Russian-backbone LAIV seed viruses are provided to some developing countries interested in the establishment their own national LAIV productions (India, Thailand, and China) through an intellectual property transfer program initiated by WHO [7]. To date, Russian-backbone 2009 pandemic and seasonal LAIVs are registered in India, and 2009 pandemic LAIV is registered in Thailand. In addition, clinical trials of the 2009 pandemic vaccine are ongoing in China.
MedImmune seasonal and pandemic LAIV viruses are currently produced by the means of reverse genetics (RG) [8]. In contrast, Russian LAIVs are being produced by classical genetic reassortment in embryonated eggs. Regardless of the preparation method, LAIV candidates retain the temperature-sensitive (ts) and coldadapted phenotypes typical for the MDVs.
Using reverse genetics to prepare LAIV viruses is restricted by the need to purchase a license from the patent holders. The RG approach allows generating reassortant viruses with any desired genomic composition by combining a set of specific plasmids encoding all necessary genes. In contrast, classical method leads to predominant selection of high-growth reassortants. As a result, vaccine candidates generated by classical technique may have higher yield in eggs than reassortants generated by reverse genetics. This advantage of classical approach is especially important for the preparation of vaccines based on pandemic viruses, which are usually grow poorly in eggs.
Taken together, reverse genetics system allows artificial manipulating with influenza virus genome and is a powerful tool to generate reassortants with desired genomic composition in vitro. However, some pandemic vaccine productions were confronted with such difficulties as the low yield of the PR8-based RG reassortants [9]. Most probably, these candidates would have grown to much higher titers if they had been produced by classical reassortment in eggs. It must be kept in mind that availability of influenza vaccines during a pandemic will largely depend on the vaccine virus yield.
Despite a number of reassortment procedure variants, a classical reassortment technique is most widely used for the development of Russian LAIV reassortants, which includes simultaneous co-infection of embryonated chicken eggs with equal infection doses of wild type parental strain and MDV (variant #1 in the Table 1). Briefly, a co-infection step is followed by 6-7 rounds of selective propagation, including three passages at the low temperature of 25-26°C. The generation and selection of reassortants are carried out in the presence of rabbit, ferret or rat anti-MDV serum treated with RDE. Cloning by endpoint dilution is performed at each of the last 3-4 passages [13,18,19].
It should be noted, that in some cases (variant #2 in the Table 1) the vaccine strains may be only prepared using the modified classical reassortment procedure [18].

Obstacles in obtaining 6:2 reassortants and some new methodological approaches for the development of Russian Pandemic Laivs
Little is known about the mechanisms underlying the reassortment of influenza virus gene segments in a cell simultaneously infected with two different viruses. Gene compatibility of the two parental viruses can limit the number of emerging reassortant variants. It was previously demonstrated that phenotypical properties of WT parental viruses might significantly influence the reassortment efficiency after co-infection with cold-adapted MDVs. Similar to virus antigenic properties, the ts phenotype of influenza A and B circulating viruses is undergoing evolutionary changes demonstrating apparently cyclic patterns: while the most of the new antigenic drift variants and all pandemic viruses are able to grow at high temperatures (temperature resistance, non-ts phenotype), the antigenically evolved WT viruses tend to become temperature sensitive [30,31].
Based on the experiences gained during the past two decades generating seasonal and pandemic LAIVs by classical reassortment technique, the highest percentage of reassortants with vaccine 6:2 genotype could be achieved when WT parental virus was resistant both to high temperatures (38-40°C) and to non-specific serum inhibitors [15,32]. For instance, a novel swineorigin influenza A/California/07/2009 (H1N1) pdm virus was found to be non-ts and inhibitor resistant, and during the process of its reassortment with Len-MDV 33 out of 34 isolated clones inherited HA an NA from WT virus and displayed 6:2 genomic composition [31].
Reassortment of Len-MDV with another swine-origin influenza A/Indiana/10/2011 (H3N2)v human virus characterized by nonts phenotype also resulted in generation of reassortants with predominantly vaccine 6:2 genotype (unpublished data).
This was in contrast with reassortment results when ts and inhibitor sensitive seasonal WT influenza viruses were crossed with ts and inhibitor resistant MDV; the number of 6:2 vaccine reassortants was very limited, while the majority of reassortants was of 7:1 genomic composition [33].
Mismatch of HA and NA of H5 avian influenza viruses during reassortment with Len-MDV was also repeatedly demonstrated for a number of H5 viruses [12,15,18,[32][33][34][35]. Of note, generation of reassortants between highly pathogenic avian influenza viruses (HPAIVs) and MDV by classical technique in embryonated chicken eggs is impossible because the HPAIVs are lethal for the embryos. For the development of Russian pandemic LAIVs against HPAIVs reassortants for inactivated vaccine subtype H5N1 on PR8 backbone (H5N1/PR8-RG) have been chosen as a source of surface proteins. The HAs of the H5N1/PR8-RG viruses were engineered to remove four basic amino acids from the HA cleavage site, resulting in viruses that were considered attenuated for birds and safe for humans. The procedure of the HA modification was described in Subbarao et al. [36]. Unfortunately, all the attempts to recover LAIV reassortant candidates bearing avian neuraminidase N1 were ineffective; no 6:2 reassortants have been generated. Reassortment of the MDV with the H5N1/PR8-RG viruses either resulted in 7:1 reassortants containing NA and six internal protein genes from Len-MDV or these parental viruses did not reassort at all. Using ultraviolet (UV) inactivation of the H5N1/PR8-RG parental viruses prior to co-infection allowed generating 7:1 reassortants on a regular basis [15,18,32,34]. Noteworthy, reassortment of low pathogenic avian influenza A/duck/Potsdam/1402-6/86 (H5N2) virus with Len-MDV also did not gain 6:2 reassortants [35].
The phenomenon of predominant inclusion of one gene over another into genome of the reassortant progeny has not been fully elucidated. One possible reason could be the selection of the most viable and high-growth variants from a mixture of all possible reassortants, which is the limitation of the classical reassortment procedure. In case of co-infection of H5 influenza viruses with Len-MDV, only 7:1 reassortants could be successfully selected. Indeed, these 7:1 reassortants grow in embryonated chicken eggs better than the corresponding 6:2 reassortants [34]. These data are in concordance with a study by Horimoto et al. [9]. The authors created reassortants between H5N1 WT and PR8 viruses and showed that the 7:1 reassortant grew significantly better than the one with 6:2 genome composition.
The influence of the length of the NA stalk on the efficiency of virus replication was revealed in several studies [37][38][39][40]

Preclinical testing of Russian Pandemic Laivs
Eight Russian pandemic LAIV candidates were prepared by classical reassortment in embryonated chicken eggs on the Len-MDV backbone [18,34,35,41,42]. Four of them had 6:2 genomic composition, and the remaining for were 7:1 genotype ( Table  2). The attenuated phenotype of the vaccine candidates was con firmed using virological methods (determination of range of temperature sensitivity and cold adaptation during reproduction in embryonated chicken eggs), molecular genetics methods (fullgenome sequencing), and in experiments on laboratory animals (mice, guinea pigs, ferrets, chicken) ( Table 3).
Vaccine viruses were considered as possessing ts phenotype if their titer in 10-11 days old embryonated chicken eggs at elevated temperatures over 39°C was ≤ 4.2 log 10 EID 50 /mL. Viruses were considered as having a ca phenotype if their titer at low temperature of 25-26°C was ≥ 5.7 log 10 EID 50 /mL [43]. The results showed that all the reassortants exhibited high reproductive activity at an optimal incubation tem per ature of 32-33°C (9.0-10.2 log 10 EID 50 /mL
The presence of all attenuating mutations described for Len-MDV [44,46]

Safety
Clinical examination of volunteers who received two doses of pandemic LAIVs during Phase I clinical trials indicated that the vaccines were well tolerated ( Table 4). No febrile reactions, no clinically significant adverse events or changes in metabolic and hematologic laboratory tests were observed after either the first or the second vaccination. The adverse events observed were limited to sore throat, fever, nasal congestion and catarrhal nasopharynx, sneezing and headache.

Vaccine virus shedding and genetic stability of LAIV isolates
The level of pandemic LAIVs' shedding detected by culturing nasal swabs in embryonated chicken eggs varied from 13.8% (A/17/mallard/Netherlands/00/95) [43] to 70.0% (A/17/duck/ Potsdam/86/92) [51] ( Table 5). All isolates were sequenced to assess the genetic stability of the vaccine virus after replication in humans. All clinical isolates were shown to preserve all attenuating mutations of the MDV. In addition, their ts/ca phenotype was tested. All isolates retained the vaccine phenotypic characteristics of cold adaptation and temperature sensitivity described for MDV [43,[51][52][53].

Immunogenicity
The immune responses to pandemic LAIVs have been extensively studied. Antibody immune responses were measured by routine hemagglutination inhibition and microneutralization tests; influenza virus-specific serum IgG and IgA antibodies, as well as local (mucosal) IgA antibodies in nasal secretions were tested by enzyme-linked immunosorbent assay. Cellular immune responses were measured by a post-vaccination increase of virus-specific CD4 + and CD8 + T-cell levels by flow cytometry cytokine assay [54]. Two doses of pandemic LAIVs induced mucosal IgA, serum HAI, IgA and IgG, and neutralizing antibodies, as well as cell-mediated immune responses in healthy adults ( Table 5). Cumulative percentage of subjects with any antibody and/or cell-mediated immune responses to pandemic LAIVs after the first and/or the second doses reached the value of 80.0-96.6% (Table 5).

Conclusion
Avian influenza viruses remain a major pandemic threat. During the development of Russian LAIV candidates against H5 HPAIVs two new methodological approaches have been employed: (i) the use of reverse genetically constructed H5N1 strains for inactivated vaccine as a source of HA and NA genes, and (ii) inactivation of wild type parental viruses by UV prior to co-infection with Len-  MDV. The use of H5N1/PR8-RG reassortants instead of wild-type HPAIVs will minimize a risk to the laboratory personnel during the process of LAIV strain preparation, and also will significantly reduce virus pathogenicity to chick embryos.
The aim of the development of LAIV candidate is to create an attenuated virus comprised of two key antigenic determinants of circulating influenza viruses, HA and NA, and six internal protein genes of the MDV, which are responsible for attenuation [55].  Genomic composition. 2 Not tested. Table 5 Cumulative data on vaccine virus shedding and immune responses to LAIV against potentially pandemic influenza viruses in vaccinated subjects after the first and/or the second doses. However, 7:1 reassortants carrying the RG-modified HA genes of H5N1 HPAIVs and the remaining genes from an attenuated MDV might be vaccine candidates of choice.
It is important to note that attenuated properties and immunogenicity of any LAIV candidate should be well balanced.
After years of large-scale studies performed by researchers from all over the world, 6:2 genomic composition has been chosen as the best ratio of wild-type and attenuated genes in the genome of vaccine reassortants [55]. Nonetheless, HA and NA genes discordance in the genome of 7:1 pandemic vaccine candidates did not affect dramatically its immunogenic properties, as compared to those of 6:2 candidates.
In spite of differences in the genomic composition (6:2 or 7:1), all eight pandemic vaccine candidates expressed phenotypic characteristics described for MDV and MDV-based seasonal LAIVs -cold adaptation, temperature sensitivity and attenuation for different animal models.
Furthermore, Phase I clinical trials of two 6:2 LAIVs and two 7:1 LAIVs completed so far demonstrated their good safety and immunogenicity profiles. Overall, an LAIV virus bearing an RG-modified HA of a HPAIV and the remaining genes from the cold-adapted MDV has several potential advantages as a pandemic vaccine candidate. Selection of such high-yield viruses for vaccine manufacture is one of the advantages of using classical reassortment method; such highyield 7:1 LAIVs against HPAIVs will stimulate pronounced antibody and cell-mediated immune responses. This approach will allow timely production of sufficient amounts of LAIVs to meet the vaccine demand during a pandemic.