In this work we have investigated biological properties of progeny hybrid viruses obtained from recombination in vitro with a transgenic MVA candidate vaccine and a naturally circulating cowpox virus. The hypothesis of this work is that the extensive use of poxvirus-vectored vaccines in future might result in natural in vivo co-infection and recombination between poxvirus-vectored vaccines and naturally circulating orthopoxviruses resulting in hybrid viruses with non-parental characteristics. MVA and a naturally circulating cowpox virus were used as parental strains to test this hypothesis in vitro. MVA is arguably the vector of choice for antigen delivery , and the wide spread use of MVA vectored vaccines (especially in wild life and domestic animals) in the future is highly likely. Cowpox virus is the ancestor of other OPVs, has broad host range, and contains the most complete repertoire of immunomodulators [32–34]. Unlike vaccinia virus where DNAemia or viremia seems to be an extremely rare event in vaccinees , DNAemia in patients with localized symptoms of cowpox virus infection seem not to be a rare event . Indeed cowpox virus DNA was detected in whole blood of two independent patients at 4 weeks post infection . Persistence of cowpox virus DNA in infected individuals increases the likelihood of recombination during co-infection with a poxvirus-vectored vaccine. Thus homologous recombination between MVA vectored vaccine and a naturally circulating cowpox virus (or other orthopoxviruses) can occur and such a recombination has the potential of generating novel hybrid viruses that will elucidate our understanding of the biology of recombinant poxviruses, as well as the putative scenarios that might arise following the release of genetically modified poxviruses in the wild.
CPXV-NOH1 underwent productive infection in all the thirteen mammalian cell lines used in this study. This is consistent with the broad host range of other cowpox virus strains [37, 38]. The progeny viruses have cell line tropism similar to CPXV-NOHI, but not to MVA-HANP. Although few host range genes have been identified in cowpox virus , sequence analysis of OPV genomes indicated that host range genes cluster at the genome termini [32, 40]. Our previous work suggested that all the progeny viruses derived their genome termini from CPXV-NOH1 . Thus the hybrid viruses might have derived their host range genes from CPXV-NOH1. With the exception of Rec 3, progeny viruses have higher levels of CPE in most mammalian cell lines than the parental strains. A possible explanation is that progeny viruses have a more effective mechanism for the shutdown of host protein synthesis.
Progeny viruses displayed plaque phenotypes different from parental strains. The plaque phenotypes of the progenies were reproducible in all the cell lines (data not shown). Large lytic plaques are often associated with efficient cell to cell spread in cell cultures while small plaques may indicate inefficient cell to cell spread [41, 42]. The genetic basis of the plaque phenotypes in the parental and progeny viruses is unknown. However, in vaccinia virus Western Reserve (VACV-WR), it has been demonstrated that five EEV proteins (gene products ofA33R,A34R,A56R,B5R,F13L) and two IEV proteins (A36R, F12L) may be involved in determining plaque phenotypes [43–49]. Although functionally intact EEV and IEV membrane proteins are associated with large plaque phenotype, they are insufficient in determining plaque size per se . It has also been reported that the production of actin tails is the major factor correlating with plaque size . Rec 2 plaques were characterized by the formation of comets. Comet formation was present, albeit to a lower degree in the parental MVA-HANP. Thus it is plausible that the genes for comet formation in Rec 2 were derived from MVA-HANP. In vaccinia virus IHD, comets are due to point mutation in theA34R open reading frame (ORF) or a second site mutation in theA33R andB5R ORFs [52, 53]. Rec 3b plaques displayed high degree of syncytium formation, a trait not observed in the parental strains and other progeny viruses. Mutation in theA56R is known to cause syncytia in vaccinia virus .
Three experimental observations made in this study have potential relevance for the release of genetically modified poxviruses into the ecosystem as well as recombination between transgenic poxviruses and naturally occurring relatives. Firstly, the transgene is deleted at high frequency in Rec 3 and MVA-HANP probably as a result of adaptation to the cell lines. Secondly, the viruses that have lost the transgene have higher virus multiplication compared to the transgene positive progenitor strain. Thirdly, there is variation in the stability of the transgene or its phenotype in different cell lines. The HA phenotype of Rec 2 was very stable in African Green Monkey derived Vero cells but became unstable in rat derived intestinal IEC-6 cells. The loss of the transgene/transgene phenotype as part of adaptation to new cells or hosts, the subsequent positive selection and accumulation of none transgene expressing virus mutants might compromise the efficacy of poxvirus vectored vaccines. The loss of the transgene will likely result in less effective vaccine since there will be less antigen to elicit robust immune responses. However since MVA undergoes abortive infection in most mammalian cell types, the accumulation of none transgene expressing viruses in the vaccinated hosts seem unlikely. The loss of the transgene or its phenotype and subsequent accumulation of none transgene expressing vector will be a likely problem for poxvirus vaccines based on replication competent vaccinia virus. The apparent variation in the stability of theHA transgene or the HA phenotype across different cell lines raises the possibility that the transgene inserted into poxviruses may have varying stability in different hosts. Although, spontaneous deletion  and truncation  of transgenes have been observed in some candidate MVA vectored vaccines, this is the first report showing that the same MVA or CPXV/MVA vectored vaccines was stable in one cell line and unstable in the other. Thus, our findings suggest that there is a host cell selection against the transgene. We have no explanation for the differential loss of the HA expression in different cell lines and among different viruses following serial passage. However, we speculate that the structure of the insertedHA transgene, the promoters driving the HA expression, the direction of the insertedHA transgene relative to the surrounding genes, the inherent stability/instability of the genetic locus at which theHA was inserted and the host cell responses to the HA protein are some of the factors that might affect the stability of theHA transgene or the loss of the HA expression following serial passage of the transgenic viruses. Genome wide mapping of these recombinant viruses will shed light on the genetic basis for the biological observations made in this study.
The purified virions of CPXV-NOH1 are brick shaped with highly corrugated surface similar to the structure of VACV-WR . The virions of MVA-HANP are pleomorphic with half of the virions being brick shaped and the other half round. The observation is in concert with a previous report . A 3D reconstruction of VACV-WR virions claimed that all IMVs are brick shaped and the observation of varied shapes in earlier studies is due to the limitation of 2D imaging . The round form observed in this study for MVA-HANP and the hybrid viruses (Figures 5B, D, E) is very close to a circle such that the difference between the length and width is 15 nm or less (Table 2). However, 2D measurements as done in this study can be affected by the angle of tilt as well as the plane at which the virions lie on the grids. Thus, our finding of spherical or round forms of virions especially in MVA need to be confirmed by 3D reconstruction of MVA and other OPV strains.
IMV is the major viral form produced in Vero cells infected with CPXV-NOH1. The dominance of IMVs over the enveloped forms may not be unconnected with the fact that CPXV-NOH1 produces V+ A-type inclusion (ATI) in infected cells (data not shown). It has been suggested that IMVs marked for sequestration into ATI may not differentiate into IEVs, CEVs or EEVs . Like CPXV-NOH1, IMVs constitute over 90% of virions produced in Vero cells infected with Rec 1 at various times post infection. However in Rec 2, the percentage of enveloped forms (IEV and CEV) is higher than that of IMV. Probably, the trans-Golgi network (TGN) wrapping and transport of IEV on microtubules is very efficient in Rec 2. Rec 3 on the other hand produces very low number of IEV or CEV and may be defective in the wrapping of IMV by the TGN. In Rec 3a and Rec 3b, the proportions of IMV and enveloped virions at 24 hpi were almost equal. Kinetics of virion formation in Rec 3a indicated high percentage of CEVs even though that of IEVs is low. It seems that some of the CEVs observed in Rec 3a infected cells were produced by plasma membrane budding of IMV rather than fusion of IEV outer envelope with the plasma membrane. Budding of IMVs through the plasma membrane has been shown to be an alternative mechanism for the production of CEVs .
The incorporation of influenza virus HA protein into the CEV and EEV of transgenic viruses has potential biosafety and immunological implications. Although foreign protein on the surface of CEVs and EEVs may enhance the humoral immune response of the host , they may also alter the host range or cell tropism of the transgenic MVA. Although MVA is host restricted and may not form sufficient CEVs/EEVs in human cells, the transgene on the MVA vector can be inserted by homologous recombination into another OPV with broad host range during mixed infection. Indeed we have shown in this work that the CEV and EEV of hybrid viruses incorporated the transgenic protein on their surface. We assume that the localization of the transgenic protein on CEV/EEV but not IMV or immature viruses is because the former derived its envelope from TGN or the plasma membrane [57, 59]. Both the TGN and plasma membrane were heavily labeled with gold particles. The cellular localization of the influenza virus transgenic protein is in agreement with other reports [60, 61]. The lack of gold particles on IMVs and immature viruses suggests that the IMV membrane is not derived from cellular membranes associated with the exocytic pathway.