Influenza Virus PB1 and Neuraminidase Gene Segments Can Cosegregate during Vaccine Reassortment Driven by Interactions in the PB1 Coding Region

ABSTRACT

Egg-grown influenza vaccine yields are maximized by infection with a seed virus produced by “classical reassortment” of a seasonal isolate with a highly egg-adapted strain. Seed viruses are selected based on a high-growth phenotype and the presence of the seasonal hemagglutinin (HA) and neuraminidase (NA) surface antigens. Retrospective analysis of H3N2 vaccine seed viruses indicated that, unlike other internal proteins that were predominantly derived from the high-growth parent A/Puerto Rico/8/34 (PR8), the polymerase subunit PB1 could be derived from either parent depending on the seasonal strain. We have recently shown that A/Udorn/307/72 (Udorn) models a seasonal isolate that yields reassortants bearing the seasonal PB1 gene. This is despite the fact that the reverse genetics-derived virus that includes Udorn PB1 with Udorn HA and NA on a PR8 background has inferior growth compared to the corresponding virus with PR8 PB1. Here we use competitive plasmid transfections to investigate the mechanisms driving selection of a less fit virus and show that the Udorn PB1 gene segment cosegregates with the Udorn NA gene segment. Analysis of chimeric PB1 genes revealed that the coselection of NA and PB1 segments was not directed through the previously identified packaging sequences but through interactions involving the internal coding region of the PB1 gene. This study identifies associations between viral genes that can direct selection in classical reassortment for vaccine production and which may also be of relevance to the gene constellations observed in past antigenic shift events where creation of a pandemic virus has involved reassortment.

IMPORTANCE Influenza vaccine must be produced and administered in a timely manner in order to provide protection during the winter season, and poor-growing vaccine seed viruses can compromise this process. To maximize vaccine yields, manufacturers create hybrid influenza viruses with gene segments encoding the surface antigens from a seasonal virus isolate, important for immunity, and others from a virus with high growth properties. This involves coinfection of cells with both parent viruses and selection of dominant progeny bearing the seasonal antigens. We show that this method of creating hybrid viruses does not necessarily select for the best yielding virus because preferential pairing of gene segments when progeny viruses are produced determines the genetic makeup of the hybrids. This not only has implications for how hybrid viruses are selected for vaccine production but also sheds light on what drives and limits hybrid gene combinations that arise in nature, leading to pandemics.

INTRODUCTION

The influenza A virus genome consists of eight negative-sense, single-stranded RNA segments (vRNA) that encode at least 10 proteins (1, 2). When a single cell is coinfected with two influenza A viruses, packaging of a mixture of segments from the two parental strains into the one virion can occur, resulting in the production of hybrid progeny (3). This process is known as reassortment (4) and, when it occurs in the natural environment, can give rise to viruses that are the precursors of pandemic strains.

Although the molecular drivers of reassortment are not yet fully understood, this process has been successfully exploited for many years to produce higher yields of egg-grown influenza vaccine. Influenza is the only disease for which a new vaccine is produced every year to accommodate the changes in the surface HA and NA antigens that arise through antigenic drift. As many as 250 million doses are produced for the northern and the southern hemisphere winter seasons (5). For both influenza A components of the seasonal vaccine (H1N1 and H3N2 subtypes), “classical reassortment” between low-yielding seasonal isolates and a highly egg-adapted virus is undertaken to improve viral growth and increase the yield of vaccine antigen per egg (3). Desirable reassortants bearing the hemagglutinin (HA) and neuraminidase (NA) of the seasonal virus are positively selected in the presence of antisera against the egg-adapted parent and subsequently cloned by limiting dilution in eggs to select the dominant and thus potentially best growing reassortant. This seed virus is then used for mass egg inoculation.

Only recently have such seed viruses been fully genotyped (6, 7). Our own retrospective analysis of past vaccine seed strains available at bioCSL, Ltd. (8), is in agreement with other similar analyses that indicate that most of the non-HA and NA genes are derived from the high-growth parent. However, the seasonal PB1 gene dominates in progeny viruses in about 50 to 60% of reassortment events. Using a model seasonal virus A/Udorn/307/72 (H3N2) (Udorn), we showed that classical reassortment with the high growth A/Puerto Rico/8/34 (H1N1) virus (PR8) exemplified the case where the reassortant progeny were predominantly viruses that included the seasonal PB1 gene. A reverse-genetics-derived Udorn PB1-containing virus with Udorn HA and NA on a PR8 background had a higher HA content per particle, but this did not completely compensate for its greatly reduced growth, which resulted in a lower overall yield of HA compared to the corresponding virus with PR8 PB1. These findings indicated that the preferential inclusion of the seasonal PB1 is not driven by selection of the fittest virus and does not necessarily result in the best virus for vaccine production (8).

In the present study, we investigated the mechanism driving selection of the less fit virus containing the seasonal PB1 gene. Competitive transfections were used to analyze the incorporation frequency of viral segments and showed that Udorn PB1 was preferentially incorporated with Udorn NA. Using chimeric PB1 genes we show that the coselection of Udorn NA and PB1 gene segments is dependent on interactions involving the central coding sequence of PB1 rather than the previously defined packaging sequences (9–12).

MATERIALS AND METHODS

Cells and media.

Human embryonic kidney (293T) cells and Madin-Darby canine kidney (MDCK) cells, obtained from an existing collection in the Department of Microbiology and Immunology, The University of Melbourne, were used in the present study. 293T cells were grown in Dulbecco modified Eagle medium (Gibco, Gaithersburg, MD) and MDCK cells grown in RPMI 1640 (Gibco). Both media were supplemented with 10% heat-inactivated fetal calf serum, 2 mM l-glutamine, 2 mM sodium pyruvate, 24 μg of gentamicin/ml, 50 μg of streptomycin/ml, and 50 IU of penicillin/ml. Cocultures of 293T cells and MDCK cells for transfection were established in Opti-MEM (Gibco) with 50 μg of streptomycin/ml and 50 IU of penicillin/ml.

Viruses and plasmids.

Individual gene segments from PR8 (Mt. Sinai lineage) and Udorn viruses were reverse transcribed and cloned into pHW2000 plasmids (St. Jude Children's Hospital, Memphis, TN). Viruses were reverse engineered using the eight-plasmid system developed by Hoffmann et al. (13). The reverse-genetics-derived viruses containing a PR8 backbone with Udorn HA, NA, PB1, or both NA and PB1 are referred to as PR8(Ud-HA), PR8(Ud-NA), PR8(Ud-PB1), and PR8(Ud-NA,PB1), respectively. Rescued viruses were amplified from transfection supernatants in fertilized hen's eggs at 10 to 12 days of embryonation. Viruses were then inoculated into multiple eggs at a constant dose of 100 PFU/egg, and allantoic fluid, sampled 3 days later, was stored separately for analysis.

Site-directed mutagenesis of the PB1-encoding plasmids to alter the packaging sequences involved the use of the QuikChange site-directed mutagenesis kit (Stratagene, Cedar Creek, TX) as previously described (14) using primers (Geneworks, Hindmarsh, South Australia, Australia) containing the intended changes (the sequences are available on request). DNA sequencing was used to confirm the presence of the desired changes.

Competitive transfections.

These were undertaken using a modified version of the eight-plasmid reverse genetics system (13). For 9- and 10-plasmid transfections, 1 μg of each plasmid was mixed with FuGene6 reagent (Roche, Penzberg, Upper Bavaria, Germany) in Opti-MEM (Gibco) and added to a coculture of 239T and MDCK cells. At 6 h posttransfection, the medium was replaced with Opti-MEM supplemented with 50 μg of streptomycin/ml and 50 IU of penicillin/ml. After 24 h, TPCK (tolylsulfonyl phenylalanyl chloromethyl ketone)-trypsin at 1 μg/ml (Worthington Biochemical Corp., Lakewood, NJ) was added, and the supernatant harvested after a further 48 h and stored at −80°C. To determine the incorporation frequencies of the gene(s) of interest, viruses in the transfection supernatant were cloned by plaque formation in MDCK cells (15). Randomly chosen plaques were picked by sampling through the agarose and resuspended into 0.05% Triton X-100. The source of the competing gene segments were identified by gene-specific reverse transcriptase PCR (RT-PCR) or sequencing. The final data are derived from three independent experiments.

Gene-specific RT-PCR.

TaqMan One-Step RT-PCR master mix kit (Applied Biosystems, Foster City, CA) was used for gene identification. Each 25-μl reaction was performed using 5 μl of plaque-picked virus suspension, 12.5 μl of 2× AmpliTaq Gold DNA polymerase mix, 0.625 μl of 40× RT enzyme mix, 1.25 μl of a 10 μM concentration of each gene-specific forward and reverse primer, and 0.25 μl of a 25 μM concentration of gene-specific probe. The RT reaction mixture was incubated for 30 min at 50°C. Amplification and detection was performed using an Applied Biosystems 7500 Fast RT-PCR system. Further details for the reaction conditions, primers (Geneworks), and TaqMan (Applied Biosystems) probe sequences are available on request.

Determination of viral growth kinetics.

Growth characteristics of viruses in MDCK cells were determined by infecting cells at a multiplicity of infection (MOI) of 0.01 PFU/cell. After 1 h absorption (at t = 0 h) the inoculum was removed, and the cells were washed and incubated at 37°C and 5% CO₂ in Opti-MEM supplemented with 50 μg of streptomycin/ml, 50 UI of penicillin/ml, and 1 μg of TPCK-trypsin (Worthington Biochemical Corp.)/ml. Cell culture supernatants were harvested at various time points postinfection and stored at −80°C for analysis. Virus titers of the supernatants were determined by plaque formation on confluent monolayers of MDCK cells.

Minigenome assay for polymerase activity.

A β-lactamase (BLA) reporter assay (16) was used to compare the intrinsic activities of viral ribonucleoprotein (RNP) complexes as previously described (8). Briefly, the pCAGGS-BLA reporter plasmid (10 ng/well), in which the BLA gene is flanked by influenza virus segment untranslated region s (UTRs; derived from H1 HA), was transfected into 293T cells, together with 10 ng of four pHW2000 plasmids, each expressing one of the three influenza virus polymerase genes (PB1, PB2, or PA) or the nucleoprotein (NP) gene. Transfected cells were incubated at 35°C and 5% CO₂. At 24 h posttransfection, the β-lactamase produced was detected, after the lysis of cells, by the addition of the LyticBLazer-FRET B/G substrate (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. The relative polymerase activity was calculated as follows: (520-nm/445-nm ratio of the sample)/(520-nm/445-nm ratio of BLA plasmid alone control).

RESULTS

A strong bias for particular gene segment constellations was observed in competitive transfections.

We have previously reported that progeny viruses resulting from reassortment of the model seasonal virus Udorn and the high-growth PR8 virus that contained Udorn HA and NA also included the seasonal PB1 gene at high frequency. This was in spite of the fact that the presence of Udorn PB1 in PR8(Ud-HA,NA,PB1) led to a virus with reduced growth compared to the corresponding virus, PR8(Ud-HA,NA), with PR8 PB1 (8). Since viral fitness did not appear to be the driver for this dominance, we investigated whether the selection of the Udorn PB1 was dependent on an interaction between Udorn PB1 and the Udorn HA and/or NA, either between the gene segments themselves or at the level of protein-protein interactions during replication. To do this, we used a 10-plasmid transfection system in which plasmids encoding all eight gene segments of PR8 were provided, together with additional Udorn PB1 and HA plasmids or Udorn PB1 and NA plasmids. The resulting viruses in the transfection supernatants were analyzed from plaque picks by RT-PCR to determine the source of the competing gene segments. Three replicate experiments were performed for each competitive transfection, and the frequencies of particular combinations of competing genes within each experiment were calculated and averaged across the three experiments (Table 1).

TABLE 1.

Frequency of particular gene combinations incorporated in progeny viruses resulting from 10-plasmid competitive transfection assays

Plasmids in competition between Udorn and PR8	Incorporated genesa	No. of viruses identified (% within expt)b			Sum of viruses (avg % ± SD)c
		Expt 1	Expt 2	Expt 3
HA and PB1	PR8 HA PR8 PB1	34 (92)	25 (71)	21 (68)	80 (77 ± 13)
	PR8 HA Udorn PB1	1 (3)	5 (14)	4 (13)	10 (10 ± 6)
	Udorn HA PR8 PB1	2 (5)	5 (14)	5 (16)	12 (12 ± 6)
	Udorn HA Udorn PB1	0 (0)	0 (0)	1 (3)	1 (1 ± 2)
NA and PB1	PR8 NA PR8 PB1	28 (64)	13 (30)	22 (42)	63 (45 ± 17)
	PR8 NA Udorn PB1	4 (9)	9 (20)	6 (12)	19 (14 ± 6)
	Udorn NA PR8 PB1	4 (9)	5 (11)	5 (10)	14 (10 ± 1)
	Udorn NA Udorn PB1	8 (18)	17 (39)	19 (37)	44 (31 ± 12)

^aCompetitive 10-plasmid transfections were established with plasmids expressing all eight PR8 gene segments plus two plasmids expressing either Udorn HA and Udorn PB1 gene segments (top part of the table) or Udorn NA and Udorn PB1 gene segments (bottom of the table). The origins of the HA and PB1 or the NA and PB1 gene segments in the progeny viruses were determined by gene-specific RT-PCR.

^bThe number of viruses with a particular gene combination is indicated, with the frequency of this gene combination in an individual experiment shown in parentheses.

^cThe total number of viruses with a given gene combination in the three experiments is indicated, with the average of the frequencies ± standard deviation in parentheses.

In the HA/PB1 competitive transfection, the combination of PR8 HA with the PR8 PB1 occurred with a frequency of 77% (Table 1), suggesting a preference for the formation of the PR8 parent virus. In classical reassortment for vaccine production, these PR8 HA-containing viruses would be removed with anti-PR8 antiserum. Of the remaining viruses that incorporated Udorn HA (13%), the strong bias for the incorporation of the PR8 PB1 gene was maintained (in 12 of 13 viruses analyzed). These data indicated that the PR8 PB1 gene was selected in the presence or absence of the Udorn HA gene, providing no evidence for a preferential incorporation of Udorn HA with Udorn PB1 genes (P = 1.00 [Fisher exact test]; for each experiment, n = 3 experiments).

In the NA/PB1 competitive transfection experiments a different bias was observed (Table 1). The wild-type constellation of PR8 NA with the PR8 PB1 was observed in about half (63 of 140) of the total viruses analyzed. Of the viruses containing the Udorn NA (58 of 140), 44 of these (76%) also contained the Udorn PB1 compared to 14 (24%) that contained PR8 PB1. These data indicate that the Udorn PB1 gene is preferentially selected in the presence of the Udorn NA gene (P < 0.05 [Fisher exact test]; for each individual experiment, n = 3).

Competitive nine-plasmid transfections were undertaken to further confirm the above observations. By competing the PR8 PB1 with Udorn PB1 on different backbone constellations (Fig. 1), we confirmed the observation from the 10-plasmid transfections. In the presence of PR8 HA and NA (Fig. 1A), the PR8 PB1 was incorporated at a significantly higher frequency than the Udorn PB1 (93% ± 12%; P = 0.0008 [Student t test], n = 3 experiments). In the presence of the Udorn HA rather than PR8 HA (Fig. 1B), the preference for the incorporation of the PR8 PB1 gene was maintained, being selected in 100% of the viruses analyzed (in all experiments, n = 3). In contrast, in the presence of the Udorn NA rather than PR8 NA (Fig. 1C), the Udorn PB1 was incorporated at a significantly higher frequency (83% ± 10%) than PR8 PB1 (P = 0.0015 [Student t test], n = 3 experiments).

FIG 1.

Frequency of competing gene combinations incorporated in progeny viruses resulting from nine-plasmid competitive transfection assays. Nine-plasmid competitive transfections were performed with a mixture of PR8 (orange) and Udorn (blue) plasmids, as represented at the top of each panel. Udorn and PR8 PB1 (A to C)- or NA (D and E)-expressing plasmids were competed on a PR8 background (A and D), a PR8(Ud-HA) background (B), a PR8(Ud-NA) background (C), and a PR8(UdPB1) background (E). The origin of the PB1 or NA gene segments from a random selection of progeny plaques was determined by gene-specific RT-PCR. The data represent the averages and standard errors of three individual experiments (each experiment analyzed between 16 and 20 plaques).

Similarly, a bias for the NA gene with the PB1 gene from the same parent was observed when PR8 NA and Udorn NA genes were competed in nine-plasmid competitive transfections (Fig. 1D and E). That is, in the presence of the PR8 PB1 (Fig. 1D), the PR8 NA was incorporated at a significantly higher frequency (97% ± 6%) than Udorn NA (P < 0.0001 [Student t test], n = 3 experiments), whereas in the presence of the Udorn PB1 (Fig. 1E), Udorn NA was preferentially incorporated (89% ± 5%) over the PR8 NA (P < 0.0001 [Student t test], n = 3 experiments). These data together strongly support a preference for matched NA and PB1 genes and suggest that the dominance of viruses containing Udorn PB1 over PR8 PB1 seen in classical reassortment of the two viruses may be driven by an interaction, at the gene or protein level, with Udorn NA and not Udorn HA.

Growth phenotypes are unable to explain the bias for incorporation of Udorn PB1 with Udorn NA genes.

Bias for the predominance of a particular gene constellation may be because it results in a fitter virus that grows more rapidly and comes to dominate the progeny. We have previously shown that the PR8(Ud-HA,NA,PB1) virus, which dominates after classical reassortment and anti-PR8 antibody selection, displays growth inferior to the corresponding virus PR8(Ud-HA,NA) with PR8 PB1 (8) and wished to determine what role fitness played in selection of the viruses produced in the present study as a result of the competition between gene segments. Eight-plasmid reverse genetics was used to construct the viruses corresponding to those produced in competitive transfections and viral growth curves assessed in MDCK cells (Fig. 2). All viruses were able to be rescued for further examination with the exception of PR8(Ud-HA,PB1). This virus was also absent from the selected progeny of the 9-plasmid competition of PR8 and Udorn PB1 on the PR8(Ud-HA) backbone (Fig. 1B) and was detected in only one of 103 plaque picks in the 10-plasmid competition of PR8 and Udorn HA and PB1 on a PR8 backbone (Table 1), suggesting that this represents an unfit gene constellation.

FIG 2.

Growth kinetics of reverse-genetics derived viruses. MDCK cells were infected with PR8 (red circles), PR8(Ud-PB1) (orange squares), PR8(Ud-HA) (yellow diamonds), PR8(Ud-NA) (green triangles), or PR8(Ud-NA,PB1) (blue triangles) at an MOI of 0.01. At the indicated time points, the amount of infectious virus released into the supernatant was determined by plaque assay. The data represent the mean and standard error of two individual experiments. Note that PR8(Ud-HA,PB1) was unable to be rescued by reverse genetics.

The inclusion of Udorn HA, Udorn NA, or both Udorn NA and Udorn PB1 on the PR8 background did not alter the replication kinetics of the viruses compared to PR8 (Fig. 2). The inclusion of Udorn PB1 on the PR8 backbone in PR8(Ud-PB1) virus however, resulted in 100-fold fewer viral progeny present at 24 h postinfection, compared to the other viruses analyzed. For reasons we cannot explain as yet, this virus showed substantial and even elevated amounts of virus after the first round of replication at 8 h, but the progeny did not appear as infectious. Overall, the large difference in the replication of PR8 and PR8(Ud-PB1) viruses indicates a much greater fitness of the wild-type PR8 virus and explains its preferential selection in 10-plasmid (Table 1) and 9-plasmid (Fig. 1A) competitive transfections. In contrast, the coselection of Udorn NA and Udorn PB1 (Table 1 and Fig. 1C) could not be explained by replication differences since PR8(Ud-NA) and PR8(Ud-NA,PB1) viruses exhibited similar growth kinetics.

The central coding region of Udorn PB1 drives its coselection with Udorn NA.

Given that the coselection of Udorn PB1, rather than PR8 PB1, with Udorn NA did not compromise fitness, we assumed that interactions at the protein level during replication of PR8(Ud-NA) and PR8(Ud-NA,PB1) viruses were most likely equally effective. We therefore examined whether selection was driven by RNA segment interactions during packaging of virions. Sequences within the PB1 UTRs and adjacent regions within the coding portion of the corresponding mRNA have been shown by deletion and mutation studies to be critical for packaging and virion formation (9–12). At the 3′ end of the PB1 vRNA, deletion of all but 30 or 33 nucleotides (nt), but not 60 or 66 nt, of the coding region adjacent to the UTR significantly reduced packaging efficiency (10), and mutational analysis showed that the greatest impact was made when the parts of the UTR or regions within the first 20 nt of the coding region were mutated (9, 12). For the present study, we have deemed the 5 nt differences between PR8 and Udorn PB1s in the first 42 nt in the coding region as relating to strain-specific packaging sequences (Fig. 3A and B). At the 5′ end, up to 40 nt of the coding region are thought to be involved in packaging (11), and there is only 1 nt change between the PB1 vRNA of the two viruses in this region (Fig. 3A and B).

FIG 3.

Packaging sequences of PR8 and Udorn PB1 vRNA and the structure of chimeric plasmids. (A) Schematic representation of PB1 terminal sequences encompassing the regions thought to be most critically involved in packaging of PB1. The 3′ and 5′ UTRs and their length in nucleotides are indicated with the terminal 12- or 13-nt sequences common to all segments shown in black. Adjacent nucleotides from the coding region of the corresponding mRNA which we deem to be part of the strain-specific packaging sequences are represented by think red lines. (B) Sequence alignment of the 3′ and 5′ packaging sequences of PR8 and Udorn PB1 vRNA. Boldface and underlined red letters indicate nucleotide differences. (C) Schematic representation of PB1 plasmid constructs produced for analysis. Orange represents PR8 sequence and blue represents Udorn sequence. The boxes at each end represent the terminal 66- and 83-nt packaging sequences.

Changes were introduced into the PR8 and Udorn PB1 plasmids by site-directed mutagenesis to create a set of chimeric PB1 genes with one or two strain-specific packaging sequences swapped between Udorn and PR8. Plasmids produced through this process are described in Fig. 3C and are designated PPU, UPP, UPU, etc., depending on the PR8 (P) or Udorn (U) origin of the 3′-packaging sequence–central coding sequence–5′-packaging sequence, respectively, when transcribed as vRNA. These were subsequently used in nine-plasmid competitive transfections and plaque picks of progeny viruses analyzed by RT-PCR or sequencing (Fig. 4, 5, and 6).

FIG 4.

Impact of Udorn-specific and PR8-specific packaging sequences on the competitive coselection of PB1 gene segments. Nine-plasmid competitive transfections were performed in which the chimeric PB1 plasmids expressing the PR8 and Udorn PB1 gene segments with swapped packaging sequences were competed as indicated in panel A above the relevant panels. These chimeric PB1 plasmids were competed on a PR8 background (B) or a PR8(Ud-NA) background (C). The origin of the PB1 or NA gene segments from a random selection of progeny plaques was determined by gene-specific RT-PCR. The data represent the average and standard error of three individual experiments (each experiment analyzed between 16 and 20 plaques).

FIG 5.

Impact of swapping packaging sequences on the cosegregation of the Udorn PB1 gene. (A) Nine-plasmid competitive transfections were performed in which the chimeric PB1 plasmids expressing Udorn PB1 with one or both of the PR8 packaging sequences were competed against the parental Udorn PB1 gene, as indicated above the relevant panels. (B and C) These chimeric PB1 plasmids were competed on a PR8 background (B) or a PR8(Ud-NA) background (C). The origin of the PB1 or NA gene segments from a random selection of progeny plaques was determined by gene-specific RT-PCR. The data represent the averages and standard errors of three individual experiments (each experiment analyzed between 15 and 20 plaques).

FIG 6.

Impact of swapping packaging sequences on the cosegregation of the PR8 PB1 gene. (A) Nine-plasmid competitive transfections were performed in which the chimeric PB1 plasmids expressing PR8 PB1 with one or both of the Udorn packaging sequences were competed against the parental PR8 PB1 gene, as indicated above the relevant panels. (B and C) These chimeric PB1 plasmids were competed on a PR8 background (B) or a PR8(Ud-NA) background (C). The origin of the PB1 or NA gene segments from a random selection of progeny plaques was determined by gene-specific RT-PCR. The data represent the averages and standard errors of three individual experiments (each experiment analyzed between 15 and 20 plaques).

Initially, we examined plasmids expressing Udorn or PR8 PB1 where the 3′ end, the 5′ end, or both ends were swapped (Fig. 4A). These plasmids were used in transfections where the remaining genes were from PR8 virus (Fig. 4B) or from PR8(Ud-NA) virus (Fig. 4C). When competed on a PR8 background (Fig. 4B), genes containing the central PR8 PB1 coding sequence, i.e., UPP, PPU, and UPU, were incorporated at a significantly higher frequency (for each, P < 0.01 [Student t test]) than those expressing the Udorn PB1 central coding sequence (PUU, UUP, or PUP), regardless of the terminal packaging sequences. In contrast, when the same chimeric PB1 genes were competed in the presence of the Udorn NA (Fig. 4C), the genes encoding the central coding sequence of Udorn PB1, i.e., PUU, UUP, and PUP, were incorporated at a significantly higher frequency (for each, P < 0.05 [Student t test]) than those containing the PR8 PB1 central coding region (UPP, PPU, or UPU), again regardless of the terminal packaging sequences. These data suggest that the central coding region of Udorn PB1 may be driving its selection over PR8 PB1 in the presence of Udorn NA.

The origin of the PB1 packaging sequences did not affect the cosegregation of Udorn PB1 with Udorn NA.

We next investigated any influence of the terminal packaging sequences in the situation where the central coding sequences of the PB1 gene were of Udorn origin in either PR8 or PR8(Ud-NA) virus backbones. Udorn PB1 plasmid was competed with Udorn plasmid having either the 3′, the 5′ or both 3′ and 5′ packaging sequences swapped for the corresponding PR8 sequence (Fig. 5A). When UUU was competed against Udorn PB1 plasmids containing the PR8-specific 3′ packaging sequences (PUU or PUP) on a PR8 background (Fig. 5B), PUU and PUP were incorporated at a significantly higher frequency than UUU (P < 0.001 [using the Student t test for each competitive transfection]). In contrast when UUU and UUP were competed, each was incorporated at equal frequencies (P = 0.0531 [Student t test]). These data suggest that the PR8-specific 3′ packaging sequence is much more important in driving the selective incorporation of Udorn PB1 into a PR8 virus than is the PR8-specific 5′ packaging sequence.

In addition, when the same plasmids were competed in the presence of the Udorn NA (Fig. 5C), no bias for terminal packaging sequences were observed (P > 0.05 [Student t test for each competitive transfection]). This finding agrees with that in Fig. 4C, again indicating that the central coding sequence of the Udorn PB1 is responsible for its coselection with Udorn NA and is irrespective of the strain specificity of the terminal packaging sequences.

The PR8-specific 3′ packaging sequence can direct preferential inclusion of PB1 genes.

The observation that the PR8-specific 3′ packaging sequence appeared important for the incorporation of a largely Udorn PB1 into a PR8 virus, prompted us to examine whether this was also true for the incorporation of PR8 PB1. Chimeric plasmids encoding the PR8 PB1 with one or both of the Udorn-specific packaging sequences (UPP, PPU, and UPU) were competed against the complete PR8 PB1 (PPP) (Fig. 6A). When PPP was competed against the PR8 PB1 gene containing the Udorn-specific 3′ packaging sequence (i.e., UPP or UPU), in the presence or absence of the Udorn NA (Fig. 5B and C), PPP was incorporated at a significantly higher frequency than UPP or UPU (P < 0.001 [using the Student t test for each competitive transfection]). Similarly to the findings described above, the single nucleotide difference in the 5′ packaging sequence did not appear to influence incorporation, with the PPP and PPU incorporated at equal frequencies (P > 0.05 [Student t test]). These data suggest that the presence of the PR8-specific PB1 3′ packaging sequence is important for the preferential incorporation of the PR8 PB1 gene segment into PR8 virus.

Full PR8 polymerase activity requires the PR8-specific PB1 3′ sequence.

The observation that the PR8-specific PB1 3′ packaging sequence is important for the preferential incorporation of the PR8 PB1 gene segment into PR8 virus does not necessarily imply a role for this region in packaging. It is also possible that the optimal functioning of the PR8 PB1 is dependent on the sequence of this region. To discriminate between these two possibilities, a minigenome assay, which does not require any packaging of virus, was performed to measure the intrinsic polymerase activities of RNP complexes comprising PR8 PB2, PA, and NP and the parental or chimeric PB1 gene segments (Fig. 7). As shown in Fig. 7A, significant differences in the function of the polymerase complexes were observed when the packaging sequences of the PR8 PB1 were altered (P = 0.0003, determined using one-way analysis of variance [ANOVA]). The inclusion of the Udorn-specific 3′ packaging sequence (UPP and UPU) resulted in a 70% reduction in the polymerase activity, which was significantly lower than PR8 PB1 (PPP) (P < 0.01 [one-way ANOVA Tukey's post test]), whereas the inclusion of the Udorn-specific 5′ packaging sequence alone (PPU) did not alter polymerase activity (P > 0.05 [one-way ANOVA Tukey's post test]). These findings were in contrast to those from experiments where the intrinsic activity of the Udorn PB1 chimeras in the context of a PR8 RNP (Fig. 7B) were measured. The presence of the wild-type Udorn PB1 (UUU) in the PR8 RNP itself resulted in an ca. 50% reduction in polymerase activity compared to the PR8 polymerase complex (data not shown), but there was no statistical difference between the different Udorn PB1s (P = 0.9 [one-way ANOVA]). These data suggest that the preferential selection of viruses with a PR8 PB1 gene segment containing the 3′ vRNA sequence of PR8 rather than Udorn is not solely due to the packaging function of this region but rather its influence on polymerase activity and therefore replicative fitness.

FIG 7.

Effect of packaging sequences on polymerase activity. Minigenome assays were performed in cells transfected with the pCAGGS-BLA reporter gene and plasmids expressing the PR8 PB2, PA and NP gene segments, together with a PB1 plasmid expressing the indicated wild-type or chimeric gene segment. (A) Polymerase activity of the RNPs containing the chimeric PB1 gene segment with the PR8 PB1 central coding region, expressed as a percentage of the activity of the wild-type PR8 RNP (PPP, orange). (B) Polymerase activity of the RNPs containing the chimeric PB1 gene segment with the Udorn PB1 central coding region, expressed as a percentage of the activity of the PR8(Ud-PB1) RNP (UUU, blue). The results represent the means and standard errors of three independent experiments.

DISCUSSION

This study was performed in order to determine the selective forces that operate during reassortment that enabled the dominance, in a model vaccine strain PR8(Ud-HA,NA), of viruses bearing the Udorn PB1 gene segment over the corresponding PR8 PB1 gene segment, despite their inferior growth. We show that the dominance of viruses with the Udorn PB1 is driven by the coselection of the PB1 gene segment with the Udorn NA gene segment during replication. Since we have previously observed (8) that ca. 50% of H3N2 strains reassorted with PR8 for vaccine seed production have seasonal PB1-containing viruses dominating the progeny, we propose that for these strains the interaction between seasonal PB1 and NA overrides other selective pressures that may be operating. Viruses with seasonal PB1 will therefore dominate, despite their decreased fitness, after selection with anti-PR8 polyclonal antiserum containing specificity for both PR8 HA and NA.

The conclusion that the seasonal PB1 was coselected with seasonal NA rather than HA was derived using competitive 9- and 10-plasmid transfections in a reverse-genetics system in which the viral progeny of such transfections could be analyzed. These experiments showed that only in the progeny where Udorn NA was present did the Udorn PB1 predominate. Examination of the effects of this coselection of Udorn NA and PB1 on viral growth showed that PR8(Ud-NA) had similar growth kinetics to PR8(Ud-NA,PB1), indicating that the coselection of the Udorn gene segments was not driven by viral fitness and therefore unlikely to involve protein-protein or protein-gene interactions that impact on replicative competency.

Interactions between pairs of plasmid-transcribed vRNAs in binding assays (17–20) and between vRNPs by electron tomography of virions (17, 21) have recently been elucidated and are postulated to direct the selective incorporation of the eight vRNA segments into virions (17–19, 21). More recently, mutations in viral gene segments that result in the disruption of intergene interactions in a binding assay have also been shown to result in increases in packaging defects when replication of viruses possessing these mutations were analyzed (20). These data together suggest that it is these RNA-RNA interactions that drive the assembly of vRNP complexes in the cytoplasm which are then transported to the plasma membrane where they are packaged into virions (22). Notably, the networks of gene segments proposed from RNA-RNA binding studies differ for the two different subtypes of virus, H3N2 and H5N2, that have been studied in this way (17–19). Potentially, this type of intergene segment interaction during viral assembly may be responsible for the coselection of seasonal NA and PB1 we observed in the present study. In fact, a consistent observation in the RNA binding studies (17, 19) was an interaction, though relatively weak, between the N2 NA and PB1 genes. For this to apply to our reassortant model, we would have to postulate that for 50% of seasonal strains, when reassorted with PR8, the PB1-NA interaction is strong enough to influence gene segment segregation, while in the other 50% of strains this is not the case. Tomography studies on gene segment interactions in virions budding from infected cells unfortunately do not discriminate between the three large polymerase gene segments and so the exquisite visual data on RNP-RNP juxtaposition and fibril bridging structures derived from these types of analyses (19, 21), though potentially more relevant than isolated RNA-RNA interactions, is not yet at a stage to inform our studies.

Through the use of site-directed mutagenesis to make the equivalent of chimeric PB1 gene segments in which PR8-specific and Udorn-specific PB1 terminal “packaging” sequences were swapped, we narrowed down the area directing the cosegregation of the Udorn PB1 and NA to the internal coding sequence of the Udorn PB1 gene. In another reported study of reassortment events the presence of incompatible terminal packaging sequences has been shown to modulate incorporation of a particular gene segment (23); however, our observation that interactions sitting outside the previously defined packaging sequences may drive coselection is supported by evidence from the above-mentioned RNA-RNA interaction studies (17, 20) in which internal sequences of certain genes could be mutated to ablate the binding interaction and a corresponding region on the interacting gene segment mutated to restore binding capacity. Furthermore, the thin fibril structures identified by electron tomography, which lie between vRNPs within virions, are thought to be RNA strands (17, 21) and are detected along the length of the segments and not merely at the end of the vRNPs where the terminal packaging sequences come together at the panhandle end.

Effects due to the sequence of the terminal packaging sequences were observed, however, in our wild-type and chimeric PB1 gene segment competitions performed as a control on a completely PR8 background rather than a PR8(Ud-NA) background. In these experiments wild-type PR8 PB1 was incorporated at a much higher frequency than the PR8 PB1 with the Udorn-specific 3′- but not 5′-terminal packaging sequence. The 3′ PB1 packaging sequence contains five synonymous nucleotide differences between PR8 and Udorn, whereas the 5′ PB1 packaging sequence only differs by a single nucleotide between the two strains and so it may not be surprising that the 3′ sequence variation had a more dramatic effect on incorporation frequency in our study. Use of a polymerase reporter assay allowed us to investigate whether a PR8-specific 3′ PB1 packaging sequence was required for optimal PR8 vRNP polymerase activity rather than for selective incorporation of the PB1 gene segment into the virion, since this assay is not dependent on virion formation. We showed that polymerase complexes containing PR8 PB1 with the Udorn-specific 3′ packaging sequences were significantly less active than those with the homologous 3′ PB1 sequence, indicating that viruses containing this chimeric PB1 may replicate less efficiently, therefore explaining their lower frequency in competitive nine-plasmid transfections compared to wild-type PR8 PB1.

We have presented data here that extend our knowledge of the mechanisms driving the genetic composition of influenza viruses. The results have revealed that the optimal incorporation of PR8 PB1 into PR8 virus is dependent on the presence of a PR8-specific 3′ packaging sequence, although the driver for incorporation is likely to be better polymerase function. We also showed that the preferential incorporation of the Udorn PB1 gene segment into a PR8 virus expressing Udorn HA and NA is driven by its interaction with the Udorn NA gene segment during assembly, most likely involving the central region of the PB1 gene segment. The latter finding, if applicable to N2-containing viruses that, when reassorted with H1N1 viruses contain the seasonal PB1, not only provides an explanation for the observed gene constellations in vaccine seeds but may also explain the introduction of the PB1, along with the HA and NA from the avian H2N2 donor when reassorting with the human H1N1 virus to produce the 1957 “Asian flu” pandemic strain (24, 25). Similarly, the NA and PB1 genes from the H3N2 donor cosegregated in the formation of the “triple reassortant” swine H1N2 virus that was a precursor to the 2009 H1N1 pandemic virus (26). The emerging picture from recent studies investigating viral packaging and reassortment is that forces that drive these processes may vary from segment to segment and virus to virus. Greater insight into the drivers of gene-segment cosegregation from studies such as this are needed to better understand not only the optimal gene constellation for vaccine production but also the evolution of influenza and the creation of novel pandemic strains.