Loss of Actin-Based Motility Impairs Ectromelia Virus Release In Vitro but Is Not Critical to Spread In Vivo

Ectromelia virus (ECTV) is an orthopoxvirus and the causative agent of mousepox. Like other poxviruses such as variola virus (agent of smallpox), monkeypox virus and vaccinia virus (the live vaccine for smallpox), ECTV promotes actin-nucleation at the surface of infected cells during virus release. Homologs of the viral protein A36 mediate this function through phosphorylation of one or two tyrosine residues that ultimately recruit the cellular Arp2/3 actin-nucleating complex. A36 also functions in the intracellular trafficking of virus mediated by kinesin-1. Here, we describe the generation of a recombinant ECTV that is specifically disrupted in actin-based motility allowing us to examine the role of this transport step in vivo for the first time. We show that actin-based motility has a critical role in promoting the release of virus from infected cells in vitro but plays a minor role in virus spread in vivo. It is likely that loss of microtubule-dependent transport is a major factor for the attenuation observed when A36R is deleted.


Introduction
Ectromelia virus (ECTV) is an orthopoxvirus endemic to mice that has been used extensively as a model for variola virus, the causative agent of smallpox [1]. In common with other orthopoxviruses, ECTV is among a select group of bacterial and viral pathogens able to harness actin-nucleating activity at the pathogen surface to promote their spread [2][3][4]. During infection of host cells, this activity provides a propulsive force that can accelerate microbes through the host cytosol (Listeria monocytogenes; Autographa californica multiple nucleopolyhedrovirus (AcMNPV)) or across the external surface of the host plasma membrane (vaccinia virus (VACV); enteropathogenic Echerichia coli (EPEC)). The contribution of actin-based motility to disease progression is difficult to assess in these complex systems due to the pleiotropy of pathogen-encoded activators of actin polymerization and the lack of availability of in vivo experimental models that faithfully reproduce endogenous infections.
Our understanding of the actin-based motility of orthopoxviruses is primarily derived from VACV, the prototypal poxvirus used as the live vaccine to eradicate smallpox. However, actin-based motility is conserved in the orthopoxvirus genus including variola virus, monkeypox virus, ECTV [5] and further afield in the poxvirus family such myxoma virus and yaba-like disease virus (YLDV) [6,7]. Replication of VACV leads to production of two mature, infectious morphological variants: mature virus (MV), with a single membrane derived from the endoplasmic reticulum; and wrapped virus (WV), which possess one or two additional membranes of a trans-Golgi network (TGN) origin [8][9][10].

Cells and Viruses
African green monkey kidney cells (BSC-1) and HeLa cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM; Invitrogen, Carlsbad, CA, USA) supplemented with 5% fetal bovine serum (FBS), 292 mg/mL L-glutamine, 100 units/mL penicillin and 100 mg/mL streptomycin (DMEM-FPSG) at 37 • C and 5% CO 2 . Viruses used in this study are ECTV-Mos (GenBank accession no. AF012825), ECTV-∆A36 which has been described previously [2], from this virus the ECTV-A36 Y112F and ECTV-A36 Res viruses were created. Briefly, plasmids containing recombination cassettes to replace the mCherry fluorescent protein and guanosine phosphotransferase (GPT) selection markers were made with the A36R gene either with the point mutation (for ECTV-A36 Y112F ) or parental sequence (for ECTV-A36 Res ) and approximately 900 bp flanking regions of homologous ECTV genomic DNA adjacent to the A36R locus. Constructs were transfected according to manufacturer's instructions (Lipofectamine 2000 reagent, Invitrogen) into HeLa cells infected with ECTV-∆A36 and homologous recombination allowed to proceed for 24 h. Cells were scraped and the resulting virus screened by loss of GPT resistance and fluorescence as described previously [22]. Viruses were purified by three rounds of plaque purification, and insertion of the correct sequence was confirmed by sequencing.

Plaque Assays
BSC-1 cells were seeded in six-well plates and grown to confluence. Viruses were diluted in serum free DMEM (SFM) and approximately 25 plaque forming units (PFU) was added to each well. After incubation at 37 • C in 5% CO 2 for 1 h, the cells were washed and overlaid with 1.5% carboxymethyl cellulose (CMC) in minimal essential medium (MEM) containing 2.5% FBS, 292 mg/mL L-glutamine, 100 units/mL penicillin and 100 mg/mL streptomycin. Cells were incubated for 6 days post infection (days pi) and then the overlay removed and cells stained with 1% crystal violet in methanol for visualization. Plaque diameter was measured by taking a straight line across the widest point of the plaque using Image J software (version 2.0.0-rc-49/1.51a, National Institutes of Health, Bethesda, MD, USA).

WV Release Assay
Six-well dishes were seeded with BSC-1 cells and incubated with virus (in triplicate) at an multiplicity of infection (MOI) of 0.1 for 1 h. Cells were then washed twice with PBS and DMEM-FPSG was added. The supernatants were collected at 24 h post infection (hpi), and spun at 10,000 rpm for 10 min at 4 • C to remove cells and cellular debris. To quantify the infectious WV, plaque assays of 10-fold serial dilutions of the supernatant were performed on BSC-1 cells as described above. After 6 days, cells were stained with methanol/1% crystal violet and plaques enumerated. All WV assays were performed on at least three separate occasions.

Ethics Statement
This study was performed in accordance with the recommendations in the Australian code of practice for the care and use of animals for scientific purposes and the Australian National Health and Medical Research Council Guidelines and Policies on Animal Ethics. The Australian National University Animal Ethics and Experimentation Committee approved all animal experiments (Protocol Numbers: J.IG.75.09 and A2012/041). Tribromoethanol (Avertin, Millipore Sigma, St. Louis, MO, USA) was used as the anesthetic (200-240 mg/kg body weight) given via intra-peritoneal injection prior to infection with virus. The respiration rate of the animals was monitored during anesthesia and recovery took place upon a warm table. Animals were euthanized by cervical dislocation. The animal ethics approval did not allow for mortality experiments and for all mice to be euthanized in the event that 20% of the mice succumbed to disease. For these reasons, all mice were euthanized at Day 5 post-infection.

Generation of ECTV-A36 Y112F
To assess the role of actin-based transport in ECTV, four viruses were used. The parental ECTV-Mos shares a high level of sequence identity with VACV Western Reserve (VACV-WR), with the exception of a large truncation at the C-terminal that removes three Asn-Pro-Phe (NPF) motifs that recruit intersectin-1 [24]. Although the WD/WE motifs that mediate kinesin-1 binding are intact, of the two critical tyrosines in VACV-WR at positions 112 and 132, only the first, Y112, is present in ECTV-Mos ( Figure 1A). The A36R gene was deleted in ECTV-Mos by replacing the gene with selectable (GPT resistance) and screenable (mCherry fluorescence) markers. This created the ∆A36 virus, which is deficient in both microtubule transport and actin-based motility [2]. From an ECTV-∆A36 background, ECTV-A36 Res and ECTV-A36R Y112F were made by rescue of the small plaque phenotype. ECTV-A36 Res restored a wild type copy of A36R and was expected to fully restore function of virus. ECTV-A36 Y112F carries a Y112F substitution, which was predicted to specifically disrupt the actin-based motility function of A36R while leaving other functions intact ( Figure 1B). Viruses were then sequenced at the A36R gene by Sanger sequencing to confirm the mutation ( Figure 1C).
Viruses 2018, 10, x 5 of 14 of the two critical tyrosines in VACV-WR at positions 112 and 132, only the first, Y112, is present in ECTV-Mos ( Figure 1A). The A36R gene was deleted in ECTV-Mos by replacing the gene with selectable (GPT resistance) and screenable (mCherry fluorescence) markers. This created the ΔA36 virus, which is deficient in both microtubule transport and actin-based motility [2]. From an ECTV-ΔA36 background, ECTV-A36 Res and ECTV-A36R Y112F were made by rescue of the small plaque phenotype. ECTV-A36 Res restored a wild type copy of A36R and was expected to fully restore function of virus. ECTV-A36 Y112F carries a Y112F substitution, which was predicted to specifically disrupt the actin-based motility function of A36R while leaving other functions intact ( Figure 1B). Viruses were then sequenced at the A36R gene by Sanger sequencing to confirm the mutation ( Figure 1C). To analyze A36 expression (the protein encoded by the A36R gene), cells were infected with ECTV-Mos, ECTV-ΔA36, ECTV-A36 Y112F and ECTV-A36 Res and the lysates collected at 48 h post infection (hpi). A36 was detected in ECTV-Mos, ECTV-A36 Y112F and ECTV-A36 Res at the expected size of 34 kDa (Figure 2A). We therefore successfully confirmed ECTV-A36 Res restores normal levels of A36 expression and that the Y112F mutation does not grossly disrupt A36 stability or mobility. We further analyzed A36 localization by immunofluorescence assay. Cells were infected with ECTV-Mos, ECTV-ΔA36, ECTV-A36 Y112F and ECTV-A36 Res and fixed at 16 hpi. Cells were stained with antibodies against A36 and known WV protein B5. As expected, A36 colocalized with B5 in ECTV-Mos, ECTV-A36 Y112F and ECTV-A36 Res , but not in ECTV-ΔA36, where it is only detected at To analyze A36 expression (the protein encoded by the A36R gene), cells were infected with ECTV-Mos, ECTV-∆A36, ECTV-A36 Y112F and ECTV-A36 Res and the lysates collected at 48 h post infection (hpi). A36 was detected in ECTV-Mos, ECTV-A36 Y112F and ECTV-A36 Res at the expected size of 34 kDa (Figure 2A). We therefore successfully confirmed ECTV-A36 Res restores normal levels of A36 expression and that the Y112F mutation does not grossly disrupt A36 stability or mobility. We further analyzed A36 localization by immunofluorescence assay. Cells were infected with ECTV-Mos, ECTV-∆A36, ECTV-A36 Y112F and ECTV-A36 Res and fixed at 16 hpi. Cells were stained with antibodies against A36 and known WV protein B5. As expected, A36 colocalized with B5 in ECTV-Mos, ECTV-A36 Y112F and ECTV-A36 Res , but not in ECTV-∆A36, where it is only detected at background levels ( Figure 2B). This confirms that both ECTV-A36 Y112F and ECTV-A36 Res viruses restore normal A36 expression and localization.

A36R Is Required for ECTV Actin-Based Motility and Virus Release
We then tested the ability of ECTV-A36 Y112F to undergo actin-based motility. HeLa cells were infected with ECTV-Mos, ECTV-∆A36, ECTV-A36 Y112F and ECTV-A36 Res and fixed at 16 hpi, then stained with anti-B5R to visualize wrapped virus and phalloidin to visualize F-actin. Actin-based motility is evident in ECTV-infected cells by the association of WV with comets of F-actin, which is absent when A36R has been deleted ( Figure 3A). Although small accumulations of F-actin could occasionally be found in association with WV in ECTV-A36 Y112F -infected cells, no comets were observed, which is consistent with an essential role for Y112F in promoting actin-based motility of ECTV. We have previously shown that release of WV by VACV is strongly attenuated by the Y112F mutation (and also by the Y112F/Y132F double mutation, VACV-A36 YdF ) [15]. In line with these results, virus release by ECTV is also dependent on actin-based motility as ECTV-A36 Y112F -infected cells displayed a dramatic (15-fold) reduction in WV release ( Figure 3B). Wild type ECTV A36R fully restored actin-based motility and WV release ( Figure 3A,B).

Actin-Based Motility Does Not Contribute to Cell-to-Cell Spread of ECTV
To determine the effect of actin-based motility on cell-to-cell spread of ECTV, plaque assays were performed in BSC-1 cells. Cells were infected with ECTV-Mos, ECTV-∆A36, ECTV-A36 Y112F and ECTV-A36 Res and plaques allowed to develop over six days. There were no significant differences in plaque size between ECTV-Mos, ECTV-A36 Y112F and ECTV-A36 Res although we were able to replicate the small plaque phenotype of ECTV-∆A36 ( Figure 4A,B). This is somewhat surprising as VACV-A36 YdF exhibits a significant reduction in cell-to-cell spread [12]. Thus, cell-to-cell spread of ECTV is primarily mediated by microtubule-dependent transport with actin-based motility playing a non-significant role.

Actin-Based Motility Plays a Minor Role in ECTV Spread In Vivo
To assess the consequences of loss of microtubule transport and actin-based motility we next examined the ability of ECTV to spread in vivo. ECTV-resistant C57BL/6 mice were infected subcutaneously in the right hind leg with 10 3 PFU virus and sacrificed five days pi, and viral dissemination to three distal sites (popliteal lymph nodes, liver and spleen) was quantified. To gain further insight into the effects of defective viral transport in vivo, three subsets of mice were used, either WT, IFN-γ −/− or IFN-α/βR −/− . Viral dissemination to the lymph nodes was reduced significantly by three orders of magnitude in ∆A36 in WT mice ( Figure 5A). However, deletion or mutation of A36 had no discernible effects on viral spread in IFN-γ −/− or IFN-α/βR −/− mice ( Figure 5B,C), with ECTV readily disseminating to the lymph nodes in these mice. In contrast, viral spread to the liver was dramatically reduced in ECTV-∆A36 and significantly reduced in ECTV A36 Y112F -infected mice ( Figure 5D-F). In all murine backgrounds, the change in viral dissemination is greatest for ECTV-∆A36 infections, while the spread of ECTV-A36 Y112F was reduced to a lesser extent. In the spleen, dissemination of ECTV-∆A36 was significantly reduced but a phenotype for ECTV-A36 Y112F was only observed in WT mice ( Figure 5G-I). These results suggest that defects in actin-based motility have a minor, but detectable, role in mediating spread of ECTV in vivo but that the primary basis for the attenuation of ECTV-∆A36 is loss of microtubule-based transport.

Discussion
Actin-based motility is a highly conserved feature of the orthopoxvirus genus. Virally induced actin nucleation has been demonstrated for VACV, ECTV, variola and monkeypox viruses [2,5,16,[25][26][27], and is predicted to be a feature of replication in other members of this genus based on the conservation of the A36R gene, with a sequence identity between 81% and 100% for all published sequences [28]. The A36 protein family includes two tyrosines embedded in a Src-family kinase substrate motif, with the exception of ECTV-A36 that contains only the N-terminal tyrosine [29]. Our results confirm that actin-based motility in ECTV is mediated by the A36 protein, and further demonstrate that the 112 tyrosine is the critical residue required for actin nucleation. This is as expected, but represents only the second instance where the basis for actin-based motility has been mapped genetically in a poxvirus [15,16,25,26]. Within the wider Poxviridae family, actin-based motility has been observed in myxomavirus and YLDV [6,7]. YLDV, a member of the yatapoxvirus genus, has been observed to form virus-associated actin tails during infection, yet lacks an obvious A36R homolog [7,30]. It was found that YLDV encodes a functional ortholog of A36R, termed YL126, which appears to promote actin-based motility. Although containing less than 15% amino acid identity to A36, it is found in an analogous genomic loci (YL124R and YL127R are orthologs of VACV A35R and A37R, respectively [28,30]), contains five tyrosines that are able to be phosphorylated, and can furthermore restore actin-based motility in ΔA36 VACV infected cells [28]. YL126 homologs have been identified in the other members of the yatapoxvirus family as well as myxomavirus, and in members of the carpripoxvirus, leporipoxvirus and suipoxvirus families [28]. The conservation of actin-based motility across diverse poxviruses suggests an important role in viral spread across a range of hosts.
Our results demonstrate that, while release of ECTV WV into the extracellular media is strongly defective in the absence of actin-based motility, loss of this motility has only minor consequences in vivo for the pathogenesis of ECTV. In contrast, the loss of the microtubule-based transport function of A36 results in a significant reduction of viral spread both in vitro and in vivo. We propose that the

Discussion
Actin-based motility is a highly conserved feature of the orthopoxvirus genus. Virally induced actin nucleation has been demonstrated for VACV, ECTV, variola and monkeypox viruses [2,5,16,[25][26][27], and is predicted to be a feature of replication in other members of this genus based on the conservation of the A36R gene, with a sequence identity between 81% and 100% for all published sequences [28]. The A36 protein family includes two tyrosines embedded in a Src-family kinase substrate motif, with the exception of ECTV-A36 that contains only the N-terminal tyrosine [29]. Our results confirm that actin-based motility in ECTV is mediated by the A36 protein, and further demonstrate that the 112 tyrosine is the critical residue required for actin nucleation. This is as expected, but represents only the second instance where the basis for actin-based motility has been mapped genetically in a poxvirus [15,16,25,26]. Within the wider Poxviridae family, actin-based motility has been observed in myxomavirus and YLDV [6,7]. YLDV, a member of the yatapoxvirus genus, has been observed to form virus-associated actin tails during infection, yet lacks an obvious A36R homolog [7,30]. It was found that YLDV encodes a functional ortholog of A36R, termed YL126, which appears to promote actin-based motility. Although containing less than 15% amino acid identity to A36, it is found in an analogous genomic loci (YL124R and YL127R are orthologs of VACV A35R and A37R, respectively [28,30]), contains five tyrosines that are able to be phosphorylated, and can furthermore restore actin-based motility in ∆A36 VACV infected cells [28]. YL126 homologs have been identified in the other members of the yatapoxvirus family as well as myxomavirus, and in members of the carpripoxvirus, leporipoxvirus and suipoxvirus families [28]. The conservation of actin-based motility across diverse poxviruses suggests an important role in viral spread across a range of hosts.
Our results demonstrate that, while release of ECTV WV into the extracellular media is strongly defective in the absence of actin-based motility, loss of this motility has only minor consequences in vivo for the pathogenesis of ECTV. In contrast, the loss of the microtubule-based transport function of A36 results in a significant reduction of viral spread both in vitro and in vivo. We propose that the critical function of A36, in terms of virulence in vivo, lies in translocating WV to the cell surface. Ablation of actin-based motility did not significantly affect the cell-to-cell spread of ECTV in vitro. This result was somewhat unexpected given that in VACV a mutation in actin-based motility via deletion of one or both tyrosines (VACV-A36 Y112F or VACV-A36 YdF ) does result in a reduction of cell-to-cell spread [15]. This suggests that though highly conserved, there is some variation among poxviruses in their requirement for actin-based motility during their replication cycle. A plaque assay quantitates the cell-to-cell spread of virus but not the contribution of spread of WV through the extracellular media, which is inhibited by the overlay. For example, treating VACV with the Abl kinase inhibitor imatinib causes substantial reduction in release of WV from infected cells, yet does not lead to a significant reduction in plaque size [31]. A possible explanation for why ablation of actin-based motility in VACV and ECTV results in different outcomes for cell-to-cell spread may be due to the repulsion of super-infecting virions. This process allows viral spread to occur faster than the rate of replication dynamics by means of extracellular WV being blocked from entry to early infected cells via cell surface signaling, promoting their spread to uninfected cells via actin-based motility [19,32]. Repulsion requires the early expression of A36 and A33 at the surface of infected cells and the expression of B5 on the surface of the repelled WV [19,32]. The absence of an observable defect in ECTV-A36 Y112F plaques could be attributable to super-repulsion not playing a role in ECTV spread. In this case, the function of actin-based motility may lie solely in promoting release. We have previously suggested that actin-based motility in poxviruses may have evolved to promote WV release; a mechanism later coopted for super-repulsion. Although our data would appear to support this hypothesis, ECTV-A36R is still an early expressed gene. Alternatively, the small plaque size of ECTV may mask the contribution of actin-based motility rendering the assay less sensitive to small differences.
Our results demonstrate only a minor role for actin-based motility for virus spread in vivo. While we observed a detectable reduction in viral titers in distal organs associated with ECTV-A36 Y112F , this was negligible in comparison to the effect of deleting A36. Our data also establish that intact IFN-γ and IFN-α/β signaling are critical for A36 function, particularly for virus dissemination to the lymph node, but less so for dissemination to the liver and spleen. While our ethics approval did not allow for mortality experiments, based on the viral load in liver and spleen of the IFN-γ −/− and IFN-α/βR −/− mice at Day 5, and based on our experience with the mousepox model, we predict that gene knock-out mice challenged with ECTV-∆A36 would have survived, whereas those challenged with ECTV-Mos, ECTV-A36 Y112F or ECTV-A36 Res would have all succumbed to mousepox. We speculate that if orders of magnitude lower doses of virus were used, we may have seen more significant differences between ECTV-Mos and ECTV-A36 Y112F or between ECTV-∆A36 and ECTV-A36 Y112F . We conclude that viral release mediated by actin-based motility is not a major determinant to the virulence of ECTV. There is little consensus in the literature as to how WV release correlates to poxvirus virulence in vivo. While poxvirus WV release is complex and not fully characterized, several mutations that increase release have been identified. However, these are typically, but not exclusively, associated with a reduction in cell-to-cell spread as measured by plaque assay and these viruses are attenuated in vivo (for example, VACV-B5P 189S ; [33][34][35][36][37]. The IHD-J strain of VACV carries a mutation that results in viral release 40 times higher than WR but forms normal sized plaques [33]. This strain displays lower mortality than WR in a murine infection model; however, when compared to other strains of VACV, there was a positive correlation of WV release with virulence. This suggests that WR, which produces low numbers of WV, may be the exception rather than the rule [38]. Conversely, a comparative study of variola virus strains found a negative correlation between WV release and virulence [39]. While its role in virulence remains cryptic, there are suggestions that high WV release may have a role in increasing the transmission of virus between individuals [38][39][40]. Future studies may examine this potential role for WV release, a context that is largely mechanistically uncharacterized.
Actin-based motility is utilized not only by poxviruses but also by a wide range of bacterial pathogens including species of Listeria, Rickettsia, Shigella, Mycobacteria, Burkholderia, and EPEC [41,42]. Understanding the molecular basis for actin-based motility can provide clues about both disease pathogenesis and cellular biology. Indeed the study of Listeria monocytogenes motility was key to identifying the Arp2/3 complex as an inducer of actin nucleation [18]. In many cases, actin-based motility of pathogens is difficult to study in whole animal models due to multiple or unclear paths of activation, activators with multiple functions, or lack of an endogenous disease model. Many of these bacteria are human pathogens, and study of their interaction with host cellular systems is hindered by species-specific interactions in small animal models. For example, multiple species-specific barriers exist when mice are infected with L. monocytogenes, such as the internalin-E-cadherin engagement during entry [17]. ActA, the nucleator of L. monocytogenes, is also highly pleiotropic with a role in the aggregation of bacteria [43]. In ECTV, we have an opportunity to study a pathogen in its natural host, whereby we can manipulate not only the gene responsible for actin-based motility but also ablate actin-based motility with the substitution of a single amino acid. Our understanding of how actin-based motility of ECTV affects the spread of virus in vivo may provide clues about disease progression among poxviruses and other pathogens, and help build a more complete picture of cytoskeletal remodelling by microbes.