We previously showed that a mutated PB1 gene improved the growth kinetics of a H3N2 influenza reassortant. Here, we showed that the same mutations improved the growth kinetics of a virus containing the A/Vietnam/1203/2004 (H5N1) haemagglutinin and neuraminidase (NA). Total protein yield and NA activity were increased when a chimeric NA was included. These increases indicated that the synergistic effect was due to the gene constellation containing both the altered PB1 gene and the chimeric NA gene.The object of this work was to improve the genomic backbone segments used in vaccine candidate viruses whilst avoiding detrimental gene constellation effects. It is not known what attributes from each segment affect the gene constellation or reassortment, but it is desirable that vaccine seed strains containing contemporary haemagglutinin (HA) and neuraminidase (NA) segments in the context of known backbone segments can be recovered. It is also desirable that those strains replicate well and that satisfactory protein yields can be harvested. We have previously shown that five mutations to the PB1 gene of the high-growth backbone virus PR/8 improved the growth kinetics of a reassortant containing the HA and NA segments from an H3N2 virus (Plant et al., 2012). These mutations (G180E, S216G, S361R, Q621R and N654S) were not selected based on any structural or functional information. The five mutations made to the PB1 gene (PB1 m5 ) were based on sequence variations found in vaccine strain reassortants. We have also described the improved growth and protein yield of a reassortant (Table 1) and all the viruses produced large plaques on MDCK cells (Fig. 1a). The longer stalk region of the H5N1 NA has been shown to contribute to pathogenicity (Zhou et al., 2009). Here, we confirmed that removing this region reduced pathogenicity, with or without the PB1 M5 allele, by determining mean death time (MDT). Virus particles (10 6 ) were inoculated into the allantoic fluid of eight eggs, which were then incubated at 35 u C and candled at 12 h intervals. Eggs were scored as alive or dead and the time of death recorded. MDT was obtained by multiplying the number of embryos dead at each interval post-inoculation by the hours post-inoculation and dividing by the total number of embryos inoculated. The MDT range was similar for all viruses (Table 1).Optimal polymerase activity is required for efficient replication (Giesendorf et al., 1986;Li et al., 2009;Ngai et al., 2013). PB1 from a seasonal strain or with targeted mutations has been shown to improve growth or HA yield of H3 reassortants (Cobbin et al., 2013;Plant et al., 2012); however, mixed results have been observed for H5 reassortants (Abt et al., 2011;Rudneva et al., 2007;Wanitchang et al., 2010). All three polymerase segments reassort together more frequently in experimental settings than natural reassortment, suggesting gene constellation effects (Macken et al., 2006;Varich et al., 2008;Wille et al., 2013). Here, we examined the growth of H5 reassortants with the PB1 m5 allele w...