The actin nucleator Spir-1 is a virus restriction factor that promotes innate immune signalling

Cellular proteins often have multiple and diverse functions. This is illustrated with protein Spir-1 that is an actin nucleator, but, as shown here, also functions to enhance innate immune signalling downstream of RNA sensing by RIG-I/MDA-5. In human and mouse cells lacking Spir-1, IRF3 and NF-κB-dependent gene activation is impaired, whereas Spir-1 overexpression enhanced IRF3 activation. Furthermore, the infectious virus titres and sizes of plaques formed by two viruses that are sensed by RIG-I, vaccinia virus (VACV) and Zika virus, are increased in Spir-1 KO cells. These observations demonstrate the biological importance of Spir-1 in the response to virus infection. Like cellular proteins, viral proteins also have multiple and diverse functions. Here, we also show that VACV virulence factor K7 binds directly to Spir-1 and that a diphenylalanine motif of Spir-1 is needed for this interaction and for Spir-1-mediated enhancement of IRF3 activation. Thus, Spir-1 is a new virus restriction factor and is targeted directly by an immunomodulatory viral protein that enhances virus virulence and diminishes the host antiviral responses.


Introduction
The host innate immune response to viral infection begins with the sensing of pathogen-associated molecular patterns (PAMPs) by pattern recognition receptors (PRRs), such as Toll-like receptors and retinoic acid-inducible gene-I (RIG-I)-like receptors (RLRs) [1,2]. The sensing of virus macromolecules, such as viral DNA or RNA, by PRRs triggers signalling cascades that culminate in the activation of the transcriptional factors interferon regulatory factor (IRF) 3, activator protein (AP) 1 and nuclear factor kappa B (NF-κB) [3]. These transcriptional factors translocate to the nucleus where they induce the transcription of genes encoding interferons (IFN), cytokines and chemokines. Once secreted from the cell, IFNs, cytokines and chemokines promote inflammation to restrict virus replication and control the infection. IFNs bind to their receptors on the surface of infected or non-infected cells, and trigger signal transduction via the JAK-STAT pathway leading to expression of interferon-stimulated genes (ISGs) that induce an antiviral state [4].
During the co-evolution with their hosts, viruses have evolved strategies to evade, suppress or exploit the host response to infection by targeting multiple steps of the host immune response [5]. Vaccinia virus (VACV), the prototypical orthopoxvirus, is well known as the live vaccine used to eradicate smallpox [6]. VACV has a large dsDNA genome of approximately 190 kbp [7], replicates in the cytoplasm [8] and encodes scores of proteins that antagonise innate immunity [9]. Interestingly, some cellular pathways, such as those leading to activation of IRF3 or NF-κB, or the JAK-STAT pathway downstream of IFNs binding to their receptors, are targeted by multiple different VACV proteins. Moreover, some of these viral antagonists are multifunctional and inhibit more than one host innate immune pathway [9].
VACV protein K7 is one such antagonist of innate immunity. K7 is a small, intracellular protein that is non-essential for virus replication in cell culture, yet contributes to virulence in both intradermal and intranasal mouse models of infection [10]. Functionally, K7 was reported to suppress NF-κB activation by binding to interleukin-1 receptor-associated kinaselike 2 (IRAK2) and tumour necrosis factor (TNF) receptor-associated factor 6 (TRAF6) [11]. K7 also inhibits IRF3 activation and binds to the DEAD-box RNA helicase 3 (DDX3) [11]. Another study reported that K7 affected regulation of histone methylation during VACV infection, by an unknown mechanism [12]. Two unbiased proteomic searches identified cellular binding partners of K7 [13,14], suggesting that K7 may have additional functions. The identification of cellular proteins targeted by viral proteins has been a useful approach to identify cellular factors that function in the recognition and restriction of virus infections [15,16]. In this study, by investigating the interaction between K7 and the cellular protein Spir-1 [14], we identified new functions for Spir-1 as an activator of innate immunity and as a restriction factor for both DNA and RNA viruses.
The protein spire homolog 1 (Spir-1, also known as SPIRE1) was first described to affect Drosophila embryogenesis [17]. Spir-1 has actin-binding domains [18] through which it nucleates actin filaments, an activity shared with the Arp2/3 complex and the formins [19]. Spir-1 is organised in multiple functional domains. The N-terminal region contains the kinase non-catalytic C-lobe domain (KIND), which mediates Spir-1 interaction with other proteins, such as the formins [20]. Spir-1 and the formins cooperate during actin nucleation [21][22][23][24][25][26][27][28][29]. The KIND domain is followed by four actin-binding Wiskott-Andrich syndrome protein homology domain 2 (WH2) domains that are responsible for actin nucleation [19,30,31]. The C-terminal region of Spir-1 contains a globular tail domain-binding motif (GTBM), which is responsible for binding to myosin V [32]. Next, there is a Spir-box (SB) domain that is conserved within the Spir protein family. Due to its similarity to a helical region of the rabphilin-3A protein that interacts with the GTPase Rab3A, it is thought that the Spir-box domain is involved in the association of Spir-1 and Rab-GTPases [24,[33][34][35]. Following the SB domain, there is a modified FYVE zinc finger domain that interacts with negatively-charged lipids in membranes. The membrane targeting specificity is mediated via the interaction of Spir-1 with other membrane-bound proteins such as the Rab-GTPases [29].
In adult mice, Spir-1 is expressed preferentially in neuronal and hematopoietic cells [36,37] and in humans the brain also has high Spir-1 expression [38]. In general, Spir-1 is involved in several actin-dependent cellular functions such as vesicle trafficking [24,34,35,39], DNA repair [40], mitochondrial division [41], and development of germ cells [17,23,42] but a role in innate immunity has not been described. Interestingly, a genome-wide association study found a single nucleotide polymorphism in Spir-1 that correlated with a different antibody response to smallpox vaccination [43].
Here, Spir-1 is shown to promote innate immune signalling downstream of dsRNA sensing. In particular, Spir-1 contributes to IRF3 activation via a diphenylalanine motif that is also necessary for the direct interaction of Spir-1 with VACV virulence factor K7. Using gain-of-function and loss-of-function cell lines, Spir-1 is shown to diminish VACV and ZIKV replication and/or spread and is therefore a virus restriction factor.
To investigate if the actin-binding domains of Spir-1 were needed for interaction with K7, two Myc-tagged truncations of Spir-1 were generated: an N-terminal region (amino acid residues 1-390), containing the four actin-binding WH2 domains and the KIND domain; and a C-terminal region, containing the SB and FYVE domains (amino acid residues 391-742, Fig  1C). These truncations, Spir-1 and Spir-2 were co-expressed with FLAG-K7 and Myc-tagged immunoprecipitation showed that the C terminus of Spir-1 was sufficient for interaction with K7, whereas the actin-binding domains were dispensable (Fig 1D, middle panel). The reciprocal IP gave the same conclusion and FLAG-K7 co-precipitated full-length Spir-1 and its C-terminal region (Fig 1D, right panel). As controls, the N-terminal region co-precipitated endogenous actin via its WH2 domains, and the C-terminal region interacted with 14-3-3, another binding partner of Spir-1 [48,49].

Ectopic Spir-1 increases IRF3-dependent gene expression
Since the interaction of Spir-1 with K7 is independent of its actin-binding domains, and K7 is a VACV immunomodulator and virulence factor, we hypothesised that Spir-1 might have an unknown function in antiviral immunity. To test this, the impact of Spir-1 on innate immune signalling pathways was assessed by luciferase reporter gene assays. First, cells were transfected with a reporter plasmid in which the expression of firefly luciferase is driven by the IFNβ promoter. IFNβ expression was induced by Sendai virus (SeV) infection, which is sensed by RIG-I [52]. Myc-Spir-1 expression alone did not affect IFNβ-dependent gene activation, however,
To map at which stage in the IRF3 pathway Spir-1 was acting, further ISG56.1 reporter assays were performed in which the pathway was stimulated by expression of its different components. Spir-1 activated the IRF3 pathway in a dose-response manner when MAVS (mitochondrial antiviral-signalling protein) was used as stimulant ( Fig 2F), but not when downstream components such as TBK1 (Fig 2G), IKKε (S1A Fig) or the constitutively active IRF3-5D ( Fig 2H) were expressed. These findings indicate Spir-1 enhances IRF3 activation at or downstream of MAVS and upstream of IKKε or TBK1. This result suggests that Spir-1 might also affect other pathways induced downstream of MAVS, such as NF-κB. To test this, NF-κB activation was assessed by reporter gene assay following RIG-I CARD stimulation in the presence of different doses of Spir-1. The lowest Spir-1 dose enhanced RIG-I/MAVS/NF-κB activation but a dose-dependent increase was not observed and higher Spir-1 doses diminished activation (S1B Fig). When combining this result with the absence of enhancement of TNF-α-induced NF-κB by Spir-1, we decided to focus on the clearer enhancement of IRF3 activation by Spir-1 (Fig 2E), as a proxy for innate immune activation. Next, N-and C-terminal fragments of Spir-1 were tested following stimulation by RIG-I CARD. Neither half of Spir-1 activated the IRF3 pathway fully, although the C-terminal region had greater activity than the N-terminal region, suggesting both halves are important (Fig 2I).

Spir-1 interaction with K7 is direct, independent of DDX3, and requires a diphenylalanine motif
K7 interacts directly with DDX3 [56,57], an adaptor protein in the IRF3 pathway [58,59]. To investigate whether K7 binding to Spir-1 was via DDX3, cells stably expressing an inducible shRNA targeting DDX3 (shDDX3) [59] were used. Knockdown, rather than knockout, of DDX3 was utilised because DDX3 encodes an essential protein [60]. Cells expressing a nonsilencing control (NSC) shRNA were used in parallel. DDX3 knockdown was induced by incubation with doxycycline (DOX) for 48 h as described [59], prior to transfection with either FLAG-tagged K7 or A49 for 24 h. Viral proteins were immunoprecipitated via their FLAG tag, followed by immunoblotting for endogenous Spir-1. DDX3 levels were reduced greatly in shDDX3 cells following DOX treatment, but not in control cells (Fig 3A). Despite this, Spir-1 co-precipitation by K7 was unaffected ( Fig 3A). A49 did not co-precipitate either DDX3 or Spir-1 ( Fig 3A). Next, the shDDX3 cells were used to determine if DDX3 contributed to Spir-1-induced IRF3 activation. After DDX3 knockdown, there was no significant difference in IRF3 stimulation by Spir-1 ( Fig 3B). Notably, K7 inhibited IRF3 activation when DDX3 was knocked down, showing K7 had another IRF3 inhibitory mechanism independent of DDX3 ( Fig 3B). Moreover, the presence of K7 reversed the activation of the pathway by Spir-1 ( Fig  3B). In summary, Spir-1 interaction with K7 and its function in the IRF3 pathway are independent of DDX3. Cell lysates were prepared and luciferase expression was measured and normalised to renilla luciferase. Data shown are representative of three independent experiments. Each independent experiment was done with a minimal of three wells for each condition and the statical analysis was done within a single experiment. Data are expressed as the mean (± SD) fold induction of the firefly luciferase activity normalised to renilla values for the stimulated versus non-stimulated samples. Immunoblots underneath each graph show the expression levels of the different proteins. The positions of molecular mass markers in kDa are shown on the right and the antibodies used are shown on the left. ns = not significant; � P < 0.05; �� P < 0.01, ���� P < 0.0001.

Spir-1 is a virus restriction factor
Since both DDX3 and Spir-1 bind to K7 and activate IRF3, their amino acid sequences were compared. The structure of K7 bound to a DDX3 peptide and subsequent structurebased mutagenesis showed that a diphenylalanine (FF) motif in DDX3 is essential for binding K7 and for IRF3 activation [57]. An alignment of the C terminus of Spir-1 and the minimal ten amino acid peptide from DDX3 (residues 81-90) needed for binding K7, showed a similar FF motif in Spir-1 ( Fig 3C). To determine if this was important for Spir-1 binding to K7, the phenylalanines were mutated to alanines (FFAA) and the mutant was expressed in cells together with FLAG-tagged K7. Wild-type (WT) Myc-Spir-1 and Myc-GFP were also expressed. Anti-Myc or anti-FLAG immunoprecipitation showed that the FFAA mutation impaired Spir-1 interaction with K7 ( Fig 3D), although it still precipitated both actin and 14-3-3, indicating normal folding. Notably, the FFAA mutant no longer activated IRF3 (Fig 3E).
To determine if the K7 interaction with Spir-1 was direct, these proteins were expressed by in vitro transcription and translation and immunoprecipitated alongside Myc-tagged TAB2 and FLAG-tagged VACV protein F12 [61] as controls. Both Spir-1 and DDX3 interacted directly with K7, whereas Spir-1 did not interact with F12, and K7 did not interact with TAB2 ( Fig 3F and 3G). Taken together, this demonstrated that Spir-1 binds directly to K7 and shares with DDX3 a conserved diphenylalanine motif, which is required for its function in the IRF3 pathway.

K7 uses the same amino acid residues to target both Spir-1 and DDX3
K7 binds to the DDX3 peptide through a hydrophobic pocket within a negatively-charged face [56,57]. Thus, interactions between K7 and the DDX3 peptide involve hydrophobic contacts, hydrogen bonds and electrostatic interactions, of which electrostatic contacts between R88 of DDX3 with both D28 and D31 of K7 are notable. D31 also forms a hydrogen bond with DDX3 S83 [57]. Using this information, K7 mutants were generated in an attempt to distinguish between K7 binding to Spir-1 and DDX3. K7 D28 and D31 were changed singly or together to alanine (D28A, D31A) and expressed alongside WT K7 or GFP, and together with Myc-tagged Spir-1 and HA-tagged DDX3. Immunoprecipitation showed that K7 D28A still bound both Spir-1 and DDX3, whilst K7 D31A had impaired binding to both proteins, especially DDX3 ( Fig 4A). Furthermore, the double mutant (DDAA) had further reduced binding to Spir-1 and no detectable binding to DDX3 ( Fig 4A). Importantly, all mutants still interacted with COPε ( Fig 4A), another K7 binding partner [13], suggesting normal K7 folding. Co-precipitation of endogenous Spir-1 with FLAG-tagged K7 and mutants gave similar conclusions ( Fig 4B). Notably, the K7 D31A mutant inhibited IRF3 activation poorly compared to the WT or D28A ( Fig 4C). Altogether, these data indicate that K7 requires D31 for interacting with both DDX3 and Spir-1 and via each interaction it inhibits the IRF3 pathway.

Innate immune responses are reduced in the absence of Spir-1
To investigate Spir-1-enhanced innate immune responses further, Spir-1 knockout (KO) HEK293T cell lines were generated by CRISPR-Cas9-mediated targeting of SPIRE1 exon 3, which is conserved in all Spir-1 isoforms (Fig 5A). After single cell selection and clonal expansion, HEK293T cell lines were confirmed to lack Spir-1 by immunoblotting ( Fig 5B) and by sequencing of both Spire-1 alleles (Fig 5A). A clone that lacked Spir-1 protein expression and in which the Spire-1 open reading frame was disrupted by frameshift mutations in both alleles, and that lacked WT sequence, was selected. Additionally, a vector expressing Myc-tagged Spir-1 was used to rescue Spir-1 expression in the KO cell line following lentiviral transduction. In parallel, both WT and KO cells were transduced with a control empty vector (EV) lentivirus. Spir-1 WT cells transduced with EV, and Spir-1 KO cells transduced with EV or Myc-Spir-1 were infected with SeV to activate the IRF3 pathway, as a proxy of innate immune responses. Supernatants were collected for ELISA and cells lysates were used for either RNA extraction or immunoblotting, which confirmed that cells did or did not express Spir-1 ( Fig  5B). The expression of Myc-Spir-1 increased in the presence of SeV, but this was not observed for endogenous Spir-1 ( Fig 5B). One possible explanation is that Myc-Spir-1 expression is driven by the human cytomegalovirus immediate early promoter [62] that contains binding sites for transcription factors such as NF-κB [63], which can also be activated during SeV infection. The phosphorylation of IRF3 at S396 (p-IRF3), a hallmark of IRF3 activation [64], increased greatly 24 h after SeV infection of WT cells (Fig 5B) but was reduced in the absence of Spir-1 and restored in the KO cells expressing Myc-Spir-1 ( Fig 5B). Moreover, after SeV infection, Spir-1 KO cells produced lower levels of mRNAs for ISG56/IFIT1 (Fig 5C), IFNB1 ( Fig 5D) and CXCL10 (Fig 5E), and secreted less CXCL-10 (Fig 5G), as measured by RT-qPCR and ELISA, respectively. Since SeV infection can also activate NF-κB downstream of dsRNA sensing, the expression of NFKBIA, an NF-κB-dependent gene, was also tested and found to be reduced in the absence of Spir-1 (Fig 5F). However, the absence of Spir-1 did not change the levels of CXCL-10 secreted in response to TNF-α, a stimulus that activates the NF-κB pathway independently of MAVS (Fig 5H).
Immortalised mouse embryonic fibroblasts (MEFs) from mice with a terminator (gene trap) between exons 3 and 4 of the Spire1 gene [65] were used to investigate the contribution

PLOS PATHOGENS
Spir-1 is a virus restriction factor of Spir-1 to innate immune activation in cells from a different mammal. After stimulation with either poly I:C or SeV infection, MEFs were lysed for immunoblotting or RNA analysis and supernatants were collected for ELISA. A reduction in p-IRF3 was seen in Spir-1 KO cells ( Fig  6A and 6B) and this was quantified (bottom graphs in Fig 6A and 6B). The Spir-1 KO MEFs also expressed less mRNAs for cytokines and chemokines upon stimulation with poly I:C or SeV infection than their WT controls (Fig 6D-6I). Both IRF3-(e.g. Ifit1) and NF-κB-(e.g. Nfkbia) dependent genes were reduced in the absence of Spir-1 upon poly I:C stimulation ( Fig  6C and 6F). Likewise, secretion of IL-6 and CXCL-10 were lower from Spir-1 KO cells (Fig 6J-6M). However, the levels of IL-6 and CXCL-10 secreted after stimulation with IL-1β were not reduced in Spir-1 KO MEFs (Fig 6N and 6O); indeed, there was a small increase in IL-6 levels in KO cells (Fig 6N). In summary, in both mouse and human cells lacking Spir-1 there is a defect in IRF3 activation. Furthermore, the NF-κB pathway activation is also affected downstream of RIG-I/MDA5, but not following stimulation with IL-1β or TNFα.

Spir-1 is a virus restriction factor
Given that Spir-1 interacts with VACV protein K7 and enhances the activation of innate immune pathways, which are critical for the host response to viral infection, the impact of Spir-1 on VACV replication and spread was assessed. Spir-1 WT and KO HEK293T cells and the derivative cell line in which Spir-1 expression was restored in the KO cells, were infected with A5-GFP-VACV, a VACV strain expressing GFP fused to capsid protein A5 [66], and 2 days (d) later plaque diameters were measured (Fig 7A and 7B). A significant increase in

PLOS PATHOGENS
Spir-1 is a virus restriction factor plaque size was seen in the absence of Spir-1 compared to both WT and rescued cells (Fig 7B), suggesting that Spir-1 restricts VACV spread and/or replication. Virus growth on those cells was also assessed. Cells were infected with 0.001 plaque-forming unit (PFU)/cell of VACV strain WR and 2 d later the virus yield was determined by plaque assay. Spir-1 KO cells yielded higher viral titres compared to both WT and Spir-1-complemented cells (Fig 7C). In parallel, the growth of a VACV strain lacking the K7L gene [10] was also assessed. Despite the absence of K7, there was still an increase in the viral titres in the Spir-1 KO cells when compared to control cell lines (Fig 7D). This can be explained by the presence of other immunomodulatory proteins encoded by VACV that inhibit either IRF3 or NF-κB activation [9].
Activation of the IRF3 and NF-κB pathways also restricts RNA viruses, therefore the impact of Spir-1 in restricting Zika virus (ZIKV), a single-stranded RNA virus, was also assessed. Spir-

PLOS PATHOGENS
Spir-1 is a virus restriction factor 1 WT, KO and Spir-1-rescued HEK293T cells were infected with ZIKV-mCherry at 0.01 PFU/ cell. Three days later, the mCherry signal was visualised by live-cell imaging and showed a greater signal in the KO cells compared to controls (Fig 7E), suggesting an increased ZIKV replication in these cells. To address this possibility, the virus yield in the supernatant of infected HEK293T cells was determined by plaque assay in VERO E6 cells (Fig 7F and 7G). Cells were infected at either 0.01 PFU/cell (Fig 7F) or 2 PFU/cell ( Fig 7G) and in the absence

PLOS PATHOGENS
Spir-1 is a virus restriction factor of Spir-1, there was an increase in ZIKV titres compared to the WT and Spir-1 restored KO cells. Collectively, loss of Spir-1 caused an increase in both virus plaque size and yield of infectious virus, showing that Spir-1 functions as a restriction factor against both a DNA (VACV) and an RNA (ZIKV) virus.

Discussion
During co-evolution with their hosts, viruses have evolved mechanisms to evade the host antiviral responses and thereby replicate and spread efficiently. Such virus proteins are numerous and have diverse functions and, therefore, can be used to gain insight into how the host innate immune system functions. VACV is a poxvirus and encodes scores of immunomodulatory proteins many of which target the innate immune system. In several cases, multiple VACV proteins engage the same pathway, but despite this, these proteins each contributes to virulence indicating non-redundant functions [9]. Sometimes a single protein can inhibit multiple pathways [9] or an open reading frame can encode multiple proteins with different functions [67]. High-throughput proteomic approaches have been used to identify binding partners of VACV proteins, or cellular proteins that are up or down-regulated during virus infection, and such cellular proteins may have antiviral activity and function in innate immunity [14,16]. For example, histone deacetylases (HDAC) 4 and 5 are each targeted for proteasomal degradation by VACV protein C6 and function as a virus restriction factor [15,16].
Here, Spir-1 is characterised as a new cellular protein that is bound by VACV protein K7 (Fig 1), a virulence factor that inhibits NF-κB and IRF3 activation [10,11]. K7 binding to Spir-1 was identified in a proteomic screen for binding partners of virus immunomodulatory proteins [14]. Here, this interaction is confirmed, demonstrated to occur at endogenous levels during VACV infection (Fig 1E), and shown to be direct (Fig 3F and 3G).
Spir-1 belongs to a family of proteins involved in actin organisation [44] but K7 interaction with Spir-1 does not require its N-terminal actin-binding domain (Fig 1D). VACV induces polymerisation of actin to facilitate virus dissemination from the cell surface and VACV mutants unable to polymerise actin spread poorly and form small plaques [68]. However, a mutant lacking K7 produces normal size plaques [10]. Given that K7 is an immunomodulatory protein, we hypothesised that Spir-1 might have a function in innate immunity. In reporter gene assays, Spir-1 over-expression did not affect JAK-STAT signalling induced by type I IFN or NF-κB activation downstream of TNFα or IL-1β (Figs 2B, 2C, 6N and 6O). However, Spir-1 enhanced the innate immune responses downstream of dsRNA sensing (Figs 2D and S1B), particularly IRF3-dependent gene expression (Fig 2D), which we explored further as a proxy for the innate immune activation. Conversely, human and murine Spir-1 knockout cell lines showed diminished IRF3 signalling, with reduced IRF3 phosphorylation and reduced transcription and expression of IRF3-responsive genes (Figs 5 and 6). Importantly, rescue of Spir-1 expression in Spir-1 KO cells restored phospho-IRF3 and cytokine expression levels (Fig 5).
To map where Spir-1 influences IRF3 activation, the pathway was activated by different agonists and the influence of Spir-1 examined. Spir-1 enhanced activation induced by SeV infection (Figs 2A, 5 and 6), poly I:C transfection (Fig 6), and overexpression of both RIG-I CARD (Fig 2D and 2E) and MAVS (Fig 2F), but not overexpression of TBK1, IKKε or a constitutively active IRF3 (Figs 2G, 2H and S1A). SeV RNA is sensed by RIG-I [69] whilst high molecular weight (HMW) poly I:C is sensed by MDA-5 [52,70]. This suggests Spir-1 acts at or downstream of MAVS, the adaptor molecule recruited downstream of RIG-I and MDA5 activation [71]. The association of RIG-I/MDA5 with MAVS results in the recruitment of tumour necrosis factor receptor-associated factors (TRAFs), such as TRAF3, leading to the phosphorylation of IRF3 by TBK1/IKKε. Other TRAFs, such as TRAF6, are also recruited, which in turn

PLOS PATHOGENS
Spir-1 is a virus restriction factor activate NF-κB via the IKK complex to promote the transcription of proinflammatory factors [72]. Thus, the expression of NF-κB-dependent genes was also reduced in the Spir-1 KO cells downstream of both SeV or poly I:C activation (Figs 5F and 6F). Collectively, these data suggest Spir-1 functions around the level of MAVS and contributes to the overall immune activation downstream of dsRNA recognition.
K7 inhibition of IRF3 activation was attributed to direct binding to DDX3, which acts as a multifunctional scaffolding adaptor culminating in IRF3 activation [58,59]. However, data presented here show K7 has additional targets within the IRF3 pathway, since Spir-1 also contributes to IRF3 activation and K7 is able to inhibit IRF3 activation even in the absence of DDX3 (Fig 3B). Furthermore, DDX3 and Spir-1 act at different positions in the pathway; DDX3 acts at the TRAF3 and TBK1/IKKε level [9,11,59,73], whilst Spir-1 acts upstream of TBK1/IKKε (Fig 2). Interestingly, both DDX3 and Spir-1 share a diphenylalanine motif that is responsible for both binding to K7 and their function in the IRF3 pathway (Fig 3) [57]. Furthermore, the same amino acid residue in K7 (D31) is crucial for its binding to both cellular proteins (Fig 4). This prevented assessment of the relative importance of K7 binding to each cellular protein. K7 has also been described as an NF-κB inhibitor downstream of TLR-and IL-1β activation [11]. However, Spir-1 does not contribute to NF-κB activation upon cytokine stimulation (Figs 2C, 5H, 6N, and 6O). Thus, it remains to be determined how K7 can interfere with NF-κB activation downstream of MAVS and if K7 interaction with Spir-1 can affect NF-κB activation.
Nonetheless, Spir-1 has an important antiviral role because cells lacking Spir-1 show enhanced VACV and ZIKV replication and/or spread (Fig 7). Comparable data for DDX3 are lacking, and obtaining such data is complicated by DDX3 being an essential protein [60]. Both K7 and Spir-1 localise in the cytoplasm [10,34]. Interestingly, mutations in the FYVE domain and in the diphenylalanine motif in Spir-1 change its sub-cellular localisation from a trans-Golgi network and post-Golgi vesicles localisation to an even cytoplasmic distribution [29,34]. Neither the N-or C-terminal halves of Spir-1 were capable of enhancing the activation the IRF3 pathway as well as full-length protein (Fig 2I), despite comparable expression levels, suggesting that both halves are important. It is possible that Spir-1 is directed to the correct location by its C-terminal domain, whist its N-terminal domain mediates Spir-1 engagement with other proteins to promote innate immune responses, although these have not been identified. This might explain why the FFAA Spir-1 mutant is no longer able to enhance activation of the IRF3 pathway. Another possibility could be that the Spir-1 binding to actin contributes to its function in promoting antiviral responses. For instance, mitochondrially-targeted β-actin affects IRF3 stabilisation and contributes to the activation of antiviral genes [74]. Whether Spir-1 is involved in this remains to be established but it could explain why the N-terminal half of Spir-1, via its actin-binding domains, contributes to IRF3 activation (Fig 2I). Regarding K7 inhibition of Spir-1 function, one possibility is that K7 may bind to the Spir-1 diphenylalanine motif and displace other proteins or sequester it and prevent it reaching its subcellular location. Spir-1 has several isoforms [45], and one, Spire-1C (also known as Spire1-E13), differs from the canonical isoform by encoding an extra exon sequence (ExonC or E13), which mediates its mitochondrial localisation and regulates mitochondrial division [41]. Even though this was not the isoform used in the present study, K7 might regulate Spir-1 trafficking to the mitochondrion, where MAVS is anchored [75].
RIG-I and MDA5 can sense RNA virus genomes directly, but also RNA generated by infection with DNA viruses [76], leading to IRF3 and NF-κB activation. Therefore, RIG-I/MDA5-triggered antiviral response is antagonised by several RNA and DNA viruses [71,77]. For instance, ZIKV has been shown to be sensed by both RIG-I and MDA-5 depending on the infected cell type (reviewed by [78]). Furthermore, it has long been known that dsRNA is

PLOS PATHOGENS
Spir-1 is a virus restriction factor generated during VACV infection [79][80][81], so it is not surprising that VACV encodes proteins to interfere with cytosolic RNA sensing. For instance, VACV protein E3 binds to dsRNA and prevents RIG-I-dependent sensing of RNA products generated via the transcription of ATrich DNA by RNA polymerase III [82,83], and the decapping enzymes D9 and D10 prevent the accumulation of dsRNA from viral complementary RNA molecules [84]. Downstream of RNA sensing, VACV encodes proteins to inhibit IRF3 activation, such as C6, which acts at TBK1/IKKε activation level [53], N2, which acts downstream of IRF3 translocation to the nucleus [55] and K7, which targets DDX3 [11]. Other VACV-encoded immunomodulatory proteins also inhibit NF-κB downstream of RNA sensing, such as B14 and A49 [9]. Here, we show Spir-1 contributes to IRF3 activation downstream of RIG-I/MDA5-dependent RNA sensing, with both Spir-1 and DDX3 sharing a similar motif which binds K7. Whether Spir-1 and DDX3 diphenylalanine also contributes to NF-κB activation remains unknown.
In summary, exploring in detail the function of VACV protein K7 led to the characterisation of an additional cellular factor, Spir-1 that promotes innate immune activation and is a virus restriction factor.

Antibodies and reagents
Primary antibodies used were from the following sources: rabbit (Rb  Table. Secondary antibodies used (1:10,000 dilution) were IRDye 680RD-conjugated goat anti-rabbit IgG or antimouse IgG and IRDye 800CW-conjugated goat anti-rabbit IgG or anti-mouse IgG (LI-COR).
Reagents used in this study were: anti-c-Myc agarose from Santa Cruz Biotechnology, and monoclonal anti-HA-agarose, clone HA-7, ANTI-FLAG M2 affinity gel and Poly-D-lysine hydrobromide (all from Sigma Aldrich). Human IFNα, human TNF-α and mouse IL-1β were from Peprotech, HMW poly(I:C) and puromycin were from InvivoGen, and doxycycline was from Melford.

Reporter gene assay
HEK293T cells were seeded in 96-well plates with 1.5 × 10 4 cells per well. After two days, cells were transfected with 60 ng per well of the firefly luciferase reporter plasmids (IFNβ, ISRE, NF-κB or ISG56.1), 10 ng per well of pTK-renilla luciferase and different amounts of the expression plasmid under test or empty vector (EV) control using polyethylenimine (PEI, CellnTec, 2 μL per 1 μg DNA). When necessary, EV plasmid was added to the transfection so that the final amount of DNA transfected was kept constant. In cases where stimulation was done by transfecting another plasmid, the same amount of EV was transfected to the nonstimulated (NS) control wells. Cells were stimulated as shown in Figs 1-7: (i) infection with SeV for 24 h (IFNβ Luc), (ii) 10 ng/mL of TNF-α for 8 h (NF-κB Luc) or (iii) with 1000 U/mL of IFNα for 8 h (ISRE-Luc). After stimulation, cells were washed with PBS, lysed with 100 uL/ well of passive lysis buffer (Promega) and firefly and renilla luciferase activities were measured using a FLUOstar luminometer (BMG). The firefly luciferase activity in each sample was normalised to the renilla luciferase activity and fold inductions were calculated relative to the nonstimulated controls for each plasmid. In all cases, data shown are representative from at least three independent experiments with at least triplicate samples analysed for each condition.

Immunoblotting
HEK293T (8 x 10 5 ) or MEFs (2 × 10 5 ) cells were seeded in 6-well plates and 24 h later cells were stimulated by infection with SeV or transfection with 5 μg/mL of poly I:C using lipofectamine 2000 (Life Technologies). After stimulation, cells were washed twice with icecold PBS, and scrapped into a cell lysis buffer containing 50 mM Tris-HCl pH 8, 150 mM NaCl, 1 mM EDTA, 10% (v/v) glycerol, 1% (v/v) Triton X-100 and 0.05% (v/v) NP-40, supplemented with protease (cOmplete Mini, Roche) and phosphatase inhibitors (PhosSTOP, Roche). Protein concentration was determined using a bicinchoninic acid protein assay kit (Pierce) before being boiled at 100˚C for 5 min. Proteins were then separated by SDS-polyacrylamide gel electrophoresis and transferred onto a nitrocellulose Amersham Protran membrane (GE Healthcare). Membranes were blocked at room temperature with either 5% (w/v) milk or 5% (w/v) bovine serum albumin (BSA, Sigma) in PBS containing 0.1% Tween 20. Then, membranes were incubated with a specific primary antibody diluted in blocking buffer at 4˚C overnight. After washing, membranes were probed with LI-COR secondary antibodies at room temperature followed by imaging using the LI-COR Odyssey imaging system, according to the manufacturer's instructions. Where indicated, protein bands from at least two independent experiments were quantified by using Odyssey software (LI-COR Biosciences).

Immunoprecipitation
HEK293T cells were seeded in 10-cm dishes (3.5 × 10 6 cells per dish) and transfected with the plasmids indicated in the Figs 1-7 using PEI. The following day, cells were washed twice with PSB, lysed in immunoprecipitation (IP) lysis buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.4, 0.5% (v/v) Nonidet P-40 (NP-40) and protease (cOmplete Mini, Roche) and phosphatase inhibitors (PhosSTOP, Roche)) and cleared by centrifugation at 21,000 g for 15 min at 4˚C. Cleared lysates were then incubated with 20 μL of ANTI-FLAG M2 Affinity Gel (Sigma Aldrich) or Anti-HA Agarose (Sigma Aldrich) for 2 h, or with 50 uL of Anti-c-Myc Agarose overnight, at 4˚C. Alternatively, proteins were expressed by in vitro transcription/translation using the TNT SP6 High-Yield Wheat Germ Protein Expression System (Promega) prior to incubation with the affinity resins. Immunoprecipitations were washed 3 or 4 times in IP buffer and bound proteins were eluted in Laemmli SDS-PAGE loading buffer and heated at 100˚C for 5 min. Samples were then analysed by SDS-PAGE and immunoblotting.

RT-qPCR
HEK293T (4 × 10 5 ) or MEFs (1 × 10 5 ) cells were seeded in 12-well plates. The next day, cells were stimulated by infection with SeV or transfection with 5 μg/mL of poly I:C using lipofectamine 2000 (Life Technologies). RNA was extracted using the RNeasy kit (QIAGEN) and 500 ng of each RNA sample was used to synthesise cDNA using Superscript III reverse transcriptase according to the manufacturer's protocol (Invitrogen). mRNA was quantified by real-time

PLOS PATHOGENS
Spir-1 is a virus restriction factor (glyceraldehyde-3-phosphate dehydrogenase) amplification from the same sample, and the fold induction of genes in stimulated samples was calculated relative to the unstimulated control. Experiments were performed in at least biological duplicate and conducted at least twice.

CRISPR/Cas9-mediated genome editing
Two guide RNAs (gRNAs) were designed using online software (http://tools.genomeengineering.org) to target SPIRE1 gene exon 3, which is shared by all Spir-1 isoforms. CRISPR/ Cas9-mediated genome editing of HEK293T cells was performed as described [89]. Briefly, px459 CRISPR/Cas9 plasmids with or without gRNA sequence were transfected into HEK293T using TransIT-LT1 transfection reagent (Mirus, MIR 2306). Puromycin (1 μg/mL) was added to transfected cells and, after 48 h, puromycin-resistant cells were serially diluted to obtain individual clones. Several clones were amplified, and a few potential knockout clones were selected by immunoblotting and confirmed by genomic DNA sequencing at the gRNA target sites. Only one gRNA was successful and one clonal cell line was confirmed to be knockout after PCRamplified genomic DNA was cloned into bacterial plasmids and multiple colonies (n = 20) were sequenced. These all contained frameshift mutations. No wild type allele was identified.

Lentivirus transductions
Lentivirus particles for transduction were generated after transient co-transfection of HEK293T cells seeded in 6-cm dishes. Cells were transfected with pCMV.dR8.91 and pMD-G vectors together with either pLDT-EV or pLDT-Myc-Spir-1. After 48 h and 72 h, the supernatant was collected and passed through a 0.45 μm filter. Lentivirus-containing supernatant was then used to infect HEK293T Spir-1 WT and KO cells. Transduced cells were selected with 1 μg/mL puromycin followed by clonal selection by serial dilution. Spir-1 expression was assessed by immunoblotting.

Virus infection
Plaque size analysis were performed in HEK293T Spir-1 WT and KO cells seeded in 6-well plates coated with poly-D-lysine (Sigma). Once confluent, cells were infected with VAC-V-A5-GFP at 20 PFU per well for 2 d. Virus plaques diameters (n = 54) were measured using AxioVision 4.8 software and a ZEISS Axio Vert.A1 fluorescent microscope. Viral replication was measured by multi-step growth analyses of HEK293T Spir-1 WT and KO cells infected with VACV-WR or VACVΔK7 at 0.001 PFU / cell or ZIKV-mCherry at 0.01 or 2 PFU/cell. For VACV, at 48 h post infection, cells were scraped in their medium and collected by centrifugation at 500 g for 5 min. Cells were subjected to three rounds of freeze-thawing before the infectious viral titre was determined by plaque assay on BSC-1 cells for 3 days. For ZIKV, at 72 h post infection, supernatants of infected cells were collected, and virus infectivity was determined by plaque assay on Vero E6 cells for 5 days. ZIKV-infected monolayers were also imaged using a Zeiss Axiovert 200M microscope. Images in Fig 7E were processed

PLOS PATHOGENS
Spir-1 is a virus restriction factor using Adobe Photoshop 2020 to enhance linearly the mCherry visualisation. BSC-1 and VERO E6 cells were then fixed with 4% paraformaldehyde (PFA) and stained with toluidine blue.

Statistical analysis
Statistical analysis was carried out using one or two-way ANOVA test where appropriate with the Bonferroni post-test, using the GraphPad Prism statistical software (Graph-Pad Software). Statistical significance is expressed as follows: ns = not significant, � P < 0.05, �� P < 0.01, ��� P < 0.001, ���� P < 0.0001.

PLOS PATHOGENS
Spir-1 is a virus restriction factor