Final reflection and potential for future research

The primary aim of this project was to identify molecular markers of adaptation of EIV to horses in the NS1 protein, the main viral antagonist of the host immune response. To aid with this investigation, a phylogenetic analysis was associated with cloning and site- directed mutagenesis techniques, as well as with reporter assays in order to identify markers of evolution of NS1 function. Then, the impact of NS1 evolutionary markers on EIV infection phenotype and virus-host interaction was evaluated using by reverse genetic systems and high-throughput sequencing technologies.

First by analysing the phylogenetic relationship of one-hundred-and-seventy-five EIV NS1 sequences (Chapter 4), and by comparing their amino acid sequences, fifteen residue changes fixed throughout evolution were identified. These changes correlated with previous reports from the literature (Murcia et al., 2011, Murcia et al., 2010, Barba and Daly, 2016). These changes comprised A112T and E186K substitutions in the early 1970s, R44K reversion and R59H, E71K, A86T and V230I substitutions in the early 1990s, and P216S substitution and C-terminal truncation (R220*) in the late 1990s. Although it was beyond the scope of this work, analysing in more detail the impact of each residue on NS1 function, as well as evaluating potential synergistic effects of these amino acids on NS1 function would be interesting.

To move forward with the characterization of EIV NS1, thirteen phylogenetically distinct NS1 proteins spanning the entire EIV lineage were cloned into an expressing vector, and tested in reporter assays (Hale et al., 2010, Kochs et al., 2007, Turkington et al., 2015) to compare their function. It was observed that during the first decade post viral emergence, NS1 strongly repressed general gene expression through a CPSF30-dependent mechanism, which required the presence of a glutamic acid at position 186. However, approximately 10 years post emergence this amino acid changed for a lysine, which abolished NS1 interaction with CPSF30 and the subsequent block of general gene expression (Chapter 4). These findings were in accordance with previous reports from the literature describing a link between viral adaption to certain host species and the loss of CPSF30 binding (Hale et al., 2010, Brown et al., 2001, Hossain et al., 2008). Interestingly, NS1 residue 186 is a direct neighbour of the highly conserved tryptophan 187, which is known to regulate NS1 effector domain (ED) dimerization (Hale et al., 2008, Kerry et al., 2011). This dimerization has been shown to reinforce NS1 anti-IFN activity by strengthening dsRNA binding (Aramini et al., 2011). More importantly NS1 ED dimerization was shown to be incompatible with CPSF30 binding (Kerry et al., 2011). Thus, it is possible that the loss of

CPSF30 binding of EIV NS1 ten years post emergence, has allowed the protein to interact with new binding partners or unmasked or reinforced important functions that were selectively advantageous for EIV, such as anti-IFN activities. In accordance with this hypothesis, transfection experiments also revealed that the repression exerted by NS1 over the IFN-b induction remained strong throughout evolution, which suggested that maintaining control over one of the main anti-viral defence of the host was selectively advantageous for the virus (Chapter 4).

Twenty years post emergence EIV NS1 was truncated of eleven-amino acids at its C- terminus, due to a non-sense mutation at codon 220. Although the effect of this truncation on NS1 function was not as marked as the E186K substitution, it seemed to affect NS1 control of general gene expression and ISG induction in transfection experiments.

In the future, it would be interesting to dissect in more details the contribution of NS1 residue 186 and C-terminal tail in the protein anti-IFN action, as well as their potential contribution to other NS1 function.

The impact of NS1 evolutionary markers, E186K and C-terminal truncation, on viral infection phenotype were then evaluated. To this end, two eight-plasmid reverse genetic systems, A/equine/Uruguay/1/1963 and A/equine/Ohio/1/2003, were used. U/63 is believed to be at the origin of the H3N8 EIV lineage and it naturally possesses a full-length NS1 protein harbouring E186. O/03 is the representative of Florida Clade 1 viruses currently circulating in the equine population, and expresses naturally a C-terminally truncated NS1 protein harbouring K186. By site-directed mutagenesis (Chapter 5), mutant versions of these two viruses for NS1 residue 186 and C-terminus were engineered. Their ability to grow and spread infection in mammalian cells were then evaluated and compared with that of their respective wild-type viruses. Surprisingly, NS1 evolutionary markers appeared to affect viral infection phenotype in a virus-context and cell type-dependent manner. For instance, the viral fitness of the contemporary virus (O/03) seemed to strongly depend on the maintenance of both NS1 evolutionary markers, K186 and C-terminal truncation. Indeed, a short NS1 with K186 was essential for O/03 replication in the presence of type I IFN, as well as to limit the establishment of an antiviral state, and delay apoptosis (Chapter 5). These characteristics are likely to be significant fitness traits, as they not only render an important arm of the host antiviral response ineffective, but they also maintain the cellular resources available for longer. Unexpectedly, NS1 K186E substitution seemed to have a high fitness cost for the contemporary virus in equine cells, particularly when introduced in the context of a short NS1 protein (Chapter 5). Indeed, the O/03-K186E mutant virus did not grow in equine cells, and did not induce protein shutdown or apoptosis, and no ISG or viral protein

could be detected during infection. Interestingly, the O/03 double mutant virus containing also the K186E substitution was not as strongly attenuated. Indeed, this virus could replicate in equine cells, produce its own viral proteins, and shutdown cellular protein production at early times post infection. Several hypotheses can be formulated to explain these discrepancies: (1) potential new functions provided by the extension of the NS1 C-terminal tail, e.g. PABPII binding or interaction with PDZ-containing proteins (Obenauer et al., 2006), could have partially compensated for the loss of other functions following the restitution of CPSF30 binding; (2) the lack of protein shutoff by O/03-K186E single mutant virus could simply be due to the inability of this virus to express its own proteins, including NS1; (3) finally, to be able to shutdown protein production in equine cells a full length NS1 protein with a glutamic acid at position 186 could be necessary. Further work would be needed to test these hypotheses.

In addition, both O/03 mutant viruses expressing a full-length NS1 protein were inducing apoptosis prematurely, and even more so when associated with E186K substitution. Accordingly, amino acids 181 to 185 have previously been involved in apoptosis delay via activation of the PI3K/Akt-pathway (Ehrhardt et al., 2006, Ehrhardt et al., 2007, Hale et al., 2006, Shin et al., 2007). However, the introduced mutations could also have affected the potential control of PKR activation (Fujimoto et al., 1998, Takizawa et al., 1996, Takizawa et al., 1995, Wada et al., 1995). Indeed, following dsRNA sensing, PKR is activated and starts a cascade of events leading to shutoff of protein production (Bergmann et al., 2000, Hatada et al., 1999) and induction of apoptosis (Takizawa et al., 1996, Van Campen et al., 1989). It would be worth testing if O/03 WT NS1 and mutant can equally interfere with PKR function or activate the PI3K pathway.

For U/63, although the effect of NS1 C-terminal truncation was negligible, the loss of CPSF30 binding (via E186K substitution) reduced significantly growth kinetics at early times post infection. Since the capacity to rapidly produce a new progeny is likely to be important for viral emergence, the glutamic acid at position 186, and subsequent ability to interact with CPSF30 must have played an important role in EIV emergence.

Interestingly, a similar evolution pattern to EIV NS1 has been observed for the North American ‘classical’ swine H1N1 lineage, whose NS segment is of avian origin. This virus maintained a full length NS1 protein until the mid-1960s, before introducing of a stop codon at position 220, as for EIV NS1, which resulted in an eleven-amino acids C-terminal truncation of the protein. These changes have subsequently been retained in the ‘classical’ swine H1N1 lineage until the present day (Hale et al., 2010).

Thus, the loss of CPSF30 binding and C-terminal truncation of NS1 seem to be a common evolutionary trait between swine and equine Influenza A viruses. Interestingly, human influenza A viruses seem somewhat different, and preferentially select for NS1 proteins binding CPSF30. For example, the NS1 protein of H5N1 viruses presented a defect in inhibiting general gene expression when the transmission occurred from birds to humans in 1997, however the viruses isolated since 1998 have gained this NS1 function (Twu et al., 2007, Clark et al., 2017). It would be interesting to compare the advantages and downsides of losing the CPSF30 binding and or the C-terminal tail in different hosts species.

To evaluate the impact of NS1 evolutionary markers on virus-host interaction the early gene expression profile of equine cells infected with wild-type and NS1 mutant viruses were analysed (Chapter 6). NS1 residue 186 and C-terminal tail were shown to affect host and viral gene expression in a virus-context dependent manner. The concomitant introduction of K186E substitution and C-terminal extension in O/03 NS1 led to an increased expression of HA, NA and NS viral genes, and decreased NP gene expression. Interestingly, previous reports from the literature showed that an increased expression of HA, NA and NS were associated with an increased replication efficiency early post infection (Park et al., 2015). Feature that was also observed for O/03-K186E-230 (Chapter 5). Although it was beyond the scope of this work, it would be of great interest to analyse in details the mechanisms behind this phenotype.

To identify possible biological interactions of cellular genes differentially expressed upon infection and identify important functional networks affected by EIV and modulated by NS1 evolutionary markers, the Ingenuity pathway analysis tool was used (Chapter 6). It was observed that the wild-type O/03 virus was able to activate a large number of cellular pathways, and notably to up-regulate cell survival pathways, such as the PI3K/AKT pathway, while down-regulating pathways associated with cell cycle arrest and cell death. More importantly, a short NS1 protein with K186 was important for these functions, as none of the O/03 mutants were able to do so.

In addition, all viruses apart from the wild-type O/03 virus were unable to control several key anti-viral players, such as IFN, IRF, or RIG-I pathways (Chapter 6), which was consistent with their attenuated phenotype in interferon competent equine cells (Chapter 5). This further emphasises the importance of NS1 evolutionary markers in EIV adaptation to horses, but also suggests that NS1 in not sufficient to improve viral fitness of an emergent virus.

Finally, some inflammatory pathways were downregulated by the wild-type O/03 virus, such as IL-17A, IL-6 or TNFR1 signalling, while others were activated, such as LPS/IL- 1, FceRI, or IL-8 signalling. More importantly, the U/63 double mutant virus positively regulated the FceRI signalling pathway as well, which has previously been associated with asthma pathogenesis (Lloyd and Marsland, 2017). This raises questions of the impact of exposure to Influenza viruses on exacerbation of chronic lung diseases.

Altogether these data suggest that the emergent EIV, which lacked the ability to control efficiently the host antiviral response, relied on a strong shutdown of general gene expression via an NS1-CPSF30-dependent mechanism. In doing so, the un-adapted virus created a short window during which it could take control of host gene expression, and produce and transmit a new progeny, before being outcompeted by the host immune response. But, this approach had strong negative impacts on host cell homeostasis, and was associated with to a premature induction of apoptosis. However, as the virus evolved in the new host, genetic mutations arose, and those that improved control of the host anti-viral responses while promoting cell survival were preferentially selected. This allowed the maintenance of the cellular resources for longer by delaying apoptosis induction, and overall increased replication efficiency (Figure 7.1).

Figure 7.1: Summary of the functional evolution of the EIV NS1 protein from viral emergence to date

A schematic representation of the molecular evolution of the NS1 protein of the H3N8 EIV during 50 years of circulation of the virus in the equine population is shown. Markers of evolution of the NS1 protein are indicated in red, and the consequences of these changes on virus-host interaction and consequences on EIV infection phenotype are indicated.

Today 1963 E 186 230 Early Premature apoptosis Limited replication Early 1970’s Interm. K 186 230 Late 1990’s Late K 186 219

Unspecific control of

gene expression

Specific control of IFN pathway

Premature apoptosis Limited replication

Delayed apoptosis Marked replication