INTRODUCTION
Influenza A virus (IAV) and influenza B virus (IBV) are members of the
Orthomyxoviridae family and have segmented genomes consisting of eight single-stranded, negative-sense viral RNA (vRNA) molecules (
1). Influenza A viruses have a broad species tropism and mainly exist in the wild aquatic fowl reservoir, whereas influenza B viruses are primarily limited to the human population, although rare infections of seals have been documented (
2–4). Despite these host reservoir differences, both influenza A and B viruses can cause severe infection in the human upper respiratory tract, leading to possible hospitalization or death, and are considered a major public health concern (
1).
Influenza A and B viruses have similar genomes that encode homologous proteins but can be distinguished by the different lengths of proteins and noncoding regions that serve as promoters for replication and transcription (
1). They are also distinguished by accessory proteins encoded from overlapping open reading frames (ORFs) and by the antigenic differences of internal proteins (
5). For instance, influenza A and B viruses both encode ion channel proteins, M2 and BM2, respectively, whereas influenza A virus expresses the PB1–F2 pathogenicity factor and influenza B virus expresses the NB ion channel, which are absent in the converse virus (
1). However, both influenza viruses encode two surface glycoproteins: hemagglutinin (HA), which is responsible for viral binding and fusion, and neuraminidase (NA), which is necessary for virus release from infected cells.
The HA and NA glycoproteins are the major antigenic determinants of influenza virus and are under immunologic pressure, leading to antigenic variants that are positively selected to avoid immune detection (
6). A drastic change in antigenicity occurs during antigenic shift, which is caused by viral genome reassortment, or the transfer of genomic segments between different viral strains in coinfected cells within an organism (
1). The antigenic diversity of the influenza virus glycoproteins is used to further classify influenza viruses, in which influenza A virus has 18 HA subtypes and 11 NA subtypes (
1,
7,
8), and influenza B virus has two lineages (the Victoria-like and Yamagata-like lineages) that are divergent from the ancestral influenza virus B/Lee/1940 (B/Lee) strain and have been cocirculating in the human population since the 1980s (
9–11). Influenza A and B viruses reassort intratypically between subtypes or lineages, but intertypic reassortment, or genetic swapping of segments between influenza A and B viruses, has never been observed (
1,
9,
12,
13).
Considering the similarities between influenza A and B viruses and their cocirculation in the human population, the lack of intertypic reassortment is somewhat surprising. The mechanism governing this segregation is unclear, but intertypic incompatibility at the protein or vRNA level has been hypothesized to be an underlying factor (
14–16). Although recombinant chimeric influenza A/B viruses have been created, suggesting that portions of influenza B virus glycoproteins can replace the function of influenza A virus glycoproteins, substantial genetic modifications of the polypeptides were required for their successful rescue (
17–19). Additionally, the RNA-dependent RNA polymerase (RdRp) of influenza A or B virus has been shown to transcribe genes flanked by the heterotypic promoter sequences (
20–22) found in the terminal noncoding regions (NCRs) of each vRNA (
23,
24). Influenza A virus vRNA can direct the incorporation of specific segments into budding virions, so that each nascent infectious virus contains a set of eight unique vRNA molecules (reviewed in reference
25). These sequence elements, referred to as packaging signals, are within the NCRs and coding regions of each influenza A viral segment and have not yet been characterized for influenza B viruses. Thus far, the mechanism governing how these sequence elements direct controlled packaging in influenza A virus and whether influenza B virus genomic packaging is directed by a similar mechanism remain unknown.
In order to gain further insight into the heterotypic incompatibilities between influenza A and B viruses, we used a single-cycle infectious influenza A virus (sciIAV) (
26) to evaluate the ability of prototypic full-length influenza B virus HAs to replace the function of the influenza A virus counterpart. Using sciIAV, we observed efficient complementation with full-length influenza B virus HA protein. Our results were corroborated when we were able to generate a single-cycle infectious influenza B virus (sciIBV) where the single-cycle backbone dictates the transgene incorporation, likely via packaging signal recognition. Lastly, we rescued three viruses consisting of an influenza A virus backbone with either the HA, NA, or HA and NA of influenza B virus. To generate these viruses, the packaging signals for the respective influenza A virus glycoproteins were appended to the ORF of influenza B virus HA or NA (
27,
28). This approach led to not only the first viable chimeric influenza A/B virus containing a full-length influenza B virus HA but also a chimeric influenza A/B virus containing full-length influenza B virus HA and NA, which retained the ability to reassort with a current circulating strain of influenza A virus.
Altogether, these results show that HA and NA of influenza A virus can be efficiently replaced by the corresponding influenza B virus glycoproteins and demonstrate that packaging signal compatibility plays an important role in the lack of intertypic reassortment, which suggests that speciation either led to or was caused by divergent packaging signal evolution.
DISCUSSION
The evolutionary divergence of type A and B influenza viruses over 4,000 years ago (
74) has been followed by further adaptation of these viruses within their natural reservoir of wild aquatic fowl and humans, respectively (
75). Through this diversification, however, both viruses have retained similar genomes, replication cycles, and general protein and enzymatic functions (
1). Despite these similarities, no reassortments between influenza A and B viruses have been described to date. Several mechanisms may prevent intertypic reassortment: inefficient protein-protein interactions, an inability of RdRps to replicate and transcribe heterotypic viral segments, specific inhibition of influenza A virus by influenza B virus NP, or, as demonstrated in this report, incompatible packaging signals that impede heterotypic gene incorporation into virions.
We demonstrate here that despite evolutionary divergence, full-length influenza B virus HA expressed by cell lines (
Fig. 1;
Table 1) can complement HA-deficient PR8 (
Fig. 2) and pH1N1 (
Fig. 3) sciIAVs at the protein level. These results are in contrast to those of three previous studies where the signal peptide (
19) or transmembrane and cytoplasmic tail domains (
18,
19,
76) of influenza A virus HA were required for the rescue of high-titer influenza A viruses expressing influenza B virus HA. In the former case, the authors suspected that packaging signals played a role because after cleavage, the signal peptide is not thought to affect the virus life cycle. In the latter case, it was hypothesized that the cytoplasmic tail or the transmembrane domain was needed to interact with the viral matrix 1 (M1) protein to facilitate virus assembly (
77). However, this hypothesis was not addressed experimentally. Although we did not evaluate whether type A or B virus HA proteins have similar binding affinities for influenza A virus M1, a sufficient interaction presumably takes place to provide comparable multicycle growth and sufficient HA pseudotyping
in vitro (
Fig. 2 and
3;
Table 1). Interestingly, these protein domains overlap the packaging signals of the HA segment, where the signal peptide (51 nt) constitutes the 3′ packaging signal and the transmembrane domain and cytoplasmic tail (108 nt) coincide with the 5′ packaging signal (
28,
63). Thus, instead of reconstituting protein-protein interactions in the chimeric influenza A/B viruses, the authors likely restored the compatibility of packaging signals for the incorporation of specific segments into budding virions.
To address if the different promoter signals found at the NCRs of segment termini account for the lack of intertypic reassortment or the ability to generate recombinant intertypic viruses, we performed minigenome assays where a viral RNA-like reporter gene was driven by the influenza A or B virus NCR promoter. Our results indicate that the RdRp of influenza A or B virus can replicate and transcribe reporter genes flanked by heterotypic NCRs, suggesting that in coinfected cells, gene segments would be replicated by both viral RdRps (
Fig. 4) (
21,
78). However, our and previous results indicate that the NP of influenza B virus inhibits type A RdRp activity (
56). This mechanism could clearly account for the lack of segment reassortment between influenza A and B viruses but does not explain why influenza A viruses carrying influenza B virus segments in the absence of influenza B virus NP cannot be generated.
Our observations left the possibility that inefficient gene incorporation restricts influenza A and B virus intertypic reassortment. Influenza A virus RNA molecules direct the specific incorporation of the eight unique vRNA segments into budding virions using packaging signals, which consist of the NCRs and nucleotides localized to the coding termini of each segment (
27,
28,
57–67). Influenza B virus likely packages genomic segments in a manner similar to that used by influenza A virus, because the nucleotides minimally required for influenza B virus HA packaging are also within the coding termini (
Fig. 5). Thus, in cells coinfected with influenza A and B viruses, the packaging signals either are incompatible or have distinct interactions that prevent intertypic packaging events. However, we have shown that appending the influenza A virus packaging signals to influenza B virus segments bypassed this regulation for both HA and NA (
Fig. 6). The mechanism by which packaging signals exert their function is not yet known, but detailed supramolecular interaction networks have been proposed on the basis of cryo-electron microscopy,
in vitro RNA-RNA interactions, and the generation of recombinant viruses (
79–81). Additionally, the authors of these studies were able to demonstrate that in the context of H3N2 and H5N2, certain segments preferentially reassort together, perhaps due to the RNA interactions found within packaging signals. Thus, it is plausible that if RNA-RNA interactions are indeed required for segment-specific packaging, then the chimeric segments that we have described herein should interact with their cognate H1N1 segments.
Current methodologies to produce influenza vaccine seed rely on either reverse genetics (
31) or classic virus reassortment (
82), where seasonal strains that contain the desired HA and NA are forced to reassort with a virus that has high rates of growth
in vitro. Our results demonstrate that viable chimeric influenza A/B viruses were immunogenic in mice (
Fig. 7;
Table 2). Additionally, our ability to reassort chimeric influenza A/B virus with currently circulating pH1N1 suggests that the appended packaging signals are sufficient to drive incorporation into new backbones (
Fig. 8). The finding that an influenza A virus containing influenza B virus HA and NA not only was immunogenic but also grew well
in vitro suggests that it is possible to develop a single-backbone vaccine platform where the virus backbone provides a high-yield phenotype and circulating influenza A or B virus HA and NA are incorporated via packaging signals. However, further research on its efficacy as an inactivated vaccine and the mutations that provide an attenuated phenotype for a live attenuated vaccine remains to be conducted.
ACKNOWLEDGMENTS
We thank Ron A. M. Fouchier (Erasmus Medical Center) for the pHW2000 Flu-GFP plasmid, Thomas M. Moran at the Center for Therapeutic Antibody Discovery at the Icahn School of Medicine at Mount Sinai for the HT103 and 15B6 monoclonal antibodies, Peter Palese and Adolfo García-Sastre (Icahn School of Medicine at Mount Sinai) for reagents, B. Paige Lawrence (University of Rochester) for the X31 virus, and John J. Treanor (University of Rochester) for anti-B/Vic and anti-B/Yam polyclonal sheep antisera. We also thank the NIAID Biodefense and Emerging Infectious Research Resources Repository (BEI Resources) for providing antibodies NR-3114, NR-3165, and NR-4540.
S.F.B. is currently supported by a University of Rochester immunology training grant (AI 007285-26). A.N. is the recipient of the University of Rochester Vaccine Fellowship (2013) and a Centers for Excellence in Influenza Research & Surveillance (CEIRS) intercollaborative training grant (2013). W.D. is currently supported by a University of Rochester pulmonary training grant (T32-HL66988 12). Research in the L.M.-S. laboratory is funded by NIH grants RO1 AI077719 and R03AI099681-01A1, the NIAID Centers of Excellence for Influenza Research and Surveillance (CEIRS HHSN266200700008C), and the University of Rochester Center for Biodefense Immune Modeling (CBIM HHSN272201000055C).