Vaccinia virus enhances its cell-to-cell spread by inducing Arp2/3-dependent actin polymerisation. This process is initiated by Src- and Abl-mediated phosphorylation of the viral transmembrane protein A36, leading to recruitment of a signalling network consisting of Grb2, Nck, WIP and N-WASP. Nck is a potent activator of N-WASP–Arp2/3-dependent actin polymerisation. However, recent observations demonstrate that an interaction between Nck and N-WASP is not required for vaccinia actin tail formation. We found that Cdc42 cooperates with Nck to promote actin tail formation by stabilising N-WASP beneath the virus. Cdc42 activation is mediated by the Rho guanine-nucleotide-exchange factor (GEF) intersectin-1 (ITSN1), which is recruited to the virus prior to its actin-based motility. Moreover, Cdc42, ITSN1 and N-WASP function collaboratively in a feed-forward loop to promote vaccinia-induced actin polymerisation. Outside the context of infection, we demonstrate that ITSN1 also functions together with Cdc42, Nck and N-WASP during phagocytosis mediated by the Fc gamma receptor. Our observations suggest that ITSN1 is an important general regulator of Cdc42-, Nck- and N-WASP-dependent actin polymerisation.
A number of intracellular pathogens have evolved mechanisms to stimulate Arp2/3-dependent actin polymerisation to promote their cell-to-cell spread (Haglund and Welch, 2011; Ireton, 2013; Welch and Way, 2013). Pathogens like Listeria monocytogenes and baculovirus promote actin-based motility by encoding proteins that directly activate the Arp2/3 complex (Haglund and Welch, 2011; Ireton, 2013). In contrast, others such as enteropathogenic Escherichia coli and vaccinia virus use more elaborate mechanisms to achieve the same goal (Campellone, 2010; Haglund and Welch, 2011; Welch and Way, 2013). It is therefore not surprising, that investigations into exactly how these different pathogens induce actin polymerisation have provided unprecedented insights into the proteins and mechanisms regulating the recruitment and activation of the Arp2/3 complex.
Vaccinia induces actin polymerisation when newly assembled virus particles leaving an infected cell fuse with and remain attached to the plasma membrane (Cudmore et al., 1996; Hollinshead et al., 2001; Ward and Moss, 2001a; Doceul et al., 2010). These adherent extracellular viruses induce an outside-in signal that locally activates Src and Abl family kinases (Frischknecht et al., 1999; Newsome et al., 2004; Reeves et al., 2005; Newsome et al., 2006). This results in the phosphorylation of tyrosine 112 and 132 of the viral transmembrane protein A36 (Frischknecht et al., 1999; Newsome et al., 2004; Newsome et al., 2006), which is polarised beneath the virus in part by adaptor protein 2 (AP-2)-mediated recruitment of clathrin following viral fusion with the plasma membrane (Humphries et al., 2012). Phosphorylated tyrosine 112 of A36 recruits Nck, which interacts with a complex of WIP and N-WASP, resulting in the activation of the Arp2/3 complex and actin polymerisation (Frischknecht et al., 1999; Moreau et al., 2000; Snapper et al., 2001; Zettl and Way, 2002; Weisswange et al., 2009; Donnelly et al., 2013). Tyrosine 132 of A36 recruits Grb2, which acts as a secondary adaptor to help stabilise Nck, WIP and N-WASP beneath the virus and promote actin polymerisation (Scaplehorn et al., 2002; Weisswange et al., 2009; Donnelly et al., 2013). Clathrin-mediated clustering of A36 also functions to stabilise N-WASP and enhance the rate of actin tail formation (Humphries et al., 2012). Ultimately, Arp2/3-induced actin polymerisation propels the virus away from the cell surface towards neighbouring cells, enhancing the spread of infection (Cudmore et al., 1995; Cudmore et al., 1996; Hollinshead et al., 2001; Ward and Moss, 2001a; Doceul et al., 2010).
Nck and N-WASP are essential for vaccinia actin tail formation, as they are required to recruit the WIP–N-WASP complex and activate the Arp2/3 complex, respectively (Frischknecht et al., 1999; Moreau et al., 2000; Snapper et al., 2001; Zettl and Way, 2002; Weisswange et al., 2009; Donnelly et al., 2013). In contrast, until recently, the role of WIP in vaccinia actin tail formation was more obscure (Donnelly et al., 2013). Previous observations outside the context of infection have shown that WIP inhibits the ability of N-WASP to activate the Arp2/3 complex until it receives an appropriate signalling input (Martinez-Quiles et al., 2001; Ho et al., 2004; Takano et al., 2008). However, we found that WIP also plays an essential role in linking Nck to N-WASP (Donnelly et al., 2013). Furthermore, dissection of the signalling network unexpectedly demonstrated that the interaction between Nck and N-WASP is not essential for vaccinia actin tail formation (Donnelly et al., 2013). This suggests that additional factors are required to regulate the ability of N-WASP to activate the Arp2/3 complex beneath the viral particle. The presence of an additional factor might also explain why the turnover rate of N-WASP beneath the virus is ∼3.5 times slower than that of Nck and WIP, even though they are both required for its recruitment (Weisswange et al., 2009; Donnelly et al., 2013).
We now demonstrate that Cdc42 is recruited to actin tails and acts to stabilise N-WASP within the complex. We further show interdependency between N-WASP and Cdc42, together with a role for the Rho guanine-nucleotide-exchange factor (GEF) ITSN1 in Cdc42 activation. Our findings with vaccinia could also be translated to phagocytosis mediated by the Fc gamma receptor, further illustrating the utility of pathogens as models to understand how signalling networks regulate Arp2/3-dependent actin polymerisation.
Cdc42 enhances vaccinia-induced actin polymerisation
Previously, we have shown that Cdc42 is recruited to vaccinia, but does not appear to play a role in promoting actin tail formation (Moreau et al., 2000). However, given that the binding of Nck to N-WASP is not essential for this process (Donnelly et al., 2013), combined with the slow turnover of N-WASP as compared to that of Nck and WIP (Weisswange et al., 2009), we decided to re-examine whether Cdc42 plays a supporting role during vaccinia actin tail formation. In agreement with Moreau et al. (Moreau et al., 2000), GFP–Cdc42 localised to the tip of vaccinia-induced actin tails beneath extracellular virions (Fig. 1A). We also found that extracellular viral particles recruited endogenous Cdc42 (Fig. 1A). To analyse the contribution of Cdc42 in actin tail formation, we infected N-WASP−/− MEFs stably expressing GFP-tagged wild-type (WT) N-WASP or the H208D mutant, which is deficient in Cdc42 binding (Miki et al., 1998; Abdul-Manan et al., 1999). We found no difference in the ability of WT and H208D N-WASP to promote actin tail formation when infected cells were assessed for the presence of at least one tail or not, as performed in Moreau et al. (Moreau et al., 2000) (Fig. 1B,C). Quantification of the number of actin tails per cell, however, revealed a significant decrease when N-WASP was unable to interact with Cdc42 (Fig. 1B,C). The number of extracellular viruses inducing actin tails was also reduced (Fig. 1C). Consistent with these observations, expression of dominant-negative GFP–Cdc42-N17 in infected HeLa cells also decreased the ability of vaccinia to induce actin tails (Fig. 1D; supplementary material Fig. S1). As Cdc42-N17 has been shown to cause effects in the absence of Cdc42 (Czuchra et al., 2005), and as the H208D mutant might also impact on interactions with the Cdc42-like GTPases TC10 or TCL (Aspenström et al., 2004), we additionally used siRNA to deplete Cdc42. Loss of Cdc42 also decreased the formation of vaccinia-induced actin tails (Fig. 1D; supplementary material Fig. S1), confirming a positive and specific role for Cdc42 during N-WASP-dependent vaccinia-induced actin polymerisation.
Cdc42 and Nck cooperate to stabilise N-WASP and promote actin tail formation
To analyze the interplay between Nck and Cdc42 in the system, we examined the consequence of the loss of their binding to N-WASP by combining the H208D mutation with N-WASP-ΔNck, which is deficient in Nck interactions (Donnelly et al., 2013). N-WASP−/− MEFs stably expressing N-WASP-ΔNck-H208D had significantly fewer actin tails than those expressing the H208D or ΔNck mutations alone, suggesting that Nck and Cdc42 act cooperatively within the vaccinia signalling network (Fig. 2A). The ability of vaccinia to induce actin tails directly impacts on the cell-to-cell spread of the virus, the efficiency of which can be assessed from the size of plaques formed in confluent cell monolayers (Ward and Moss, 2001b; Doceul et al., 2010; Humphries et al., 2012). As expected, in the absence of N-WASP and actin tail formation, the plaques were very small because viral cell-to-cell spread is inefficient (Fig. 2B). Stable expression of GFP–N-WASP in N-WASP−/− cells significantly enhanced viral spread. The N-WASP-H208D mutant was less effective in promoting viral spread, and the combined mutation (ΔNck-H208D) was even more defective, consistent with their reduced ability to support actin tail formation (Fig. 2A,B).
Our previous observations have shown that binding of Grb2 to phosphorylated tyrosine 132 of A36 helps to stabilise N-WASP beneath the virus and promote actin polymerisation (Scaplehorn et al., 2002; Weisswange et al., 2009; Donnelly et al., 2013). Given this, we examined whether Grb2 is responsible for the remaining actin tails seen in N-WASP−/− cells expressing GFP–N-WASP-ΔNck-H208D. We found that the proportion of cells infected with vaccinia expressing A36-Y132F with one or more actin tails was significantly reduced when N-WASP cannot bind Nck and Cdc42 (Fig. 2D). The number of actin tails was also diminished compared to that induced by the WT virus (strain WR) (Fig. 1C; Fig. 2D). Furthermore, the actin structures induced by the A36-Y132F virus resembled puncta rather than an elongated tail (Fig. 2C). Further to this, analysis of the speed of viral actin-based movements showed a significant decrease in motility for both the H208D mutation and the combined ΔNck H208D mutant in both WR and A36-Y132F infections (supplementary material Fig. S2A). Taken together, these results highlight the importance of Cdc42 in actin tail formation.
To examine whether the changes in actin tail formation reflect underlying differences in the ability of Cdc42 to stabilise N-WASP, we performed fluorescence recovery after photobleaching (FRAP) experiments (Fig. 2E). Consistent with our previous observations, we found that the half-life of recovery for GFP–N-WASP is 2.72±0.20 seconds (±s.e.m.) (Weisswange et al., 2009; Humphries et al., 2012; Donnelly et al., 2013). We recently demonstrated that the loss of Nck binding results in a significant, albeit modest, increase in the rate of GFP–N-WASP-ΔNck turnover (2.41±0.13 as compared to 2.94±0.17 seconds) (Donnelly et al., 2013). In contrast, we found that an inability to interact with Cdc42 results in a dramatic increase in the rate of GFP–N-WASP-H208D exchange, with the half-life of recovery dropping to 1.18±0.10 seconds (Fig. 2E). Addition of the ΔNck mutation further decreased the GFP–N-WASP-ΔNck-H208D half-life to 0.86±0.06 seconds (Fig. 2E). To further confirm the importance of Cdc42, we also performed FRAP experiments on Cdc42-depleted cells. In the absence of Cdc42, the turnover of N-WASP was again significantly enhanced as the half-life of recovery changed from 2.67±0.19 to 1.40±0.12 seconds (supplementary material Fig. S2B). Our observations demonstrate that the interaction with both Cdc42 and Nck acts to stabilise N-WASP within the signalling network. However, it is Cdc42 that plays the major role in stabilising N-WASP to promote actin tail formation.
Cdc42 and N-WASP act in a feed-forward loop
To investigate further the relationship between Cdc42 and N-WASP, we took advantage of the WH1-CRIB domains of N-WASP, which unlike the full-length protein, are unable to interact with Nck or activate the Arp2/3 complex. Consequently, WH1-CRIB acts as a dominant negative for actin tail formation when expressed in infected cells, as it competes with endogenous N-WASP for WIP binding and recruitment to the virus (Fig. 3A) (Moreau et al., 2000). In contrast, we found that GFP–WH1-CRIB containing the H208D mutation was far less effective at blocking actin tail formation, consistent with its weaker recruitment and a more prominent presence of endogenous N-WASP on the virus (Fig. 3A). This could suggest that Cdc42 participates in N-WASP recruitment, but it is more likely to reflect the inability of GFP–WH1-CRIB-H208D to be stabilised beneath the virus after its recruitment.
We next took advantage of N-WASP−/− MEFs to determine whether Cdc42 acts upstream of N-WASP in the vaccinia signalling network. We found that, in the absence of N-WASP, there was little recruitment of GFP–Cdc42 to virus particles (Fig. 3B). Consistent with this, live imaging revealed that the level of GFP–Cdc42 recruited to virus particles increases during actin tail initiation before reaching a relatively constant value (Fig. 3C; supplementary material Movie 1). Analysing the intensity of GFP-tagged Nck and N-WASP on the virus revealed that their levels also increase during actin tail formation (Fig. 3C; supplementary material Movies 2, 3). Moreover, Nck and N-WASP are recruited to the virus slightly before Cdc42. However, once Cdc42 is recruited, its intensity increases at a similar rate to that of Nck and N-WASP, with all signals plateauing at a similar time (Fig. 3C). This suggests that after its initial Nck-dependent recruitment, mediated by WIP, N-WASP interacts with Cdc42 to maintain sustained actin polymerisation.
ITSN-1L promotes actin tail formation
To further investigate the interdependency between Cdc42 and N-WASP, we sought to determine how Cdc42 is activated by vaccinia. As vaccinia transiently recruits clathrin in an AP-2-dependent manner prior to actin tail formation (Humphries et al., 2012), we examined whether ITSN1, an endocytic protein and Cdc42 GEF that also interacts with N-WASP (Hussain et al., 2001; Hunter et al., 2013), is recruited to the virus. Consistent with its role in endocytosis, immunofluorescence analysis revealed ITSN1 is present in numerous punctate structures (Fig. 4A). However, in vaccinia-infected cells, we additionally found that endogenous ITSN1 associated with extracellular viruses inducing actin tails (Fig. 4A). The ITSN1 antibody detects both ITSN-1L (long) and ITSN-1S (short), the latter of which has no GEF activity because it lacks the C-terminal DH-PH domains found in ITSN-1L (Hussain et al., 1999; Hussain et al., 2001). We found that GFP-tagged ITSN-1L and ITSN-1S are both present on the tips of actin tails (Fig. 4A). ITSN1 also colocalises with AP-2-associated virus particles, suggesting that ITSN1 is recruited upstream of actin tail formation at the point of clathrin accumulation (Fig. 4A). Consistent with this, ITSN1 is recruited to the virus in N-WASP−/− MEFs even in the absence of robust Cdc42 recruitment (Fig. 4B). RNAi-mediated loss of AP-2, however, did not impact on the localisation of ITSN1 on the virus, suggesting that its recruitment is independent of AP-2 and clathrin (Fig. 4C).
To determine the functional relevance of ITSN1 in actin tail formation, we used RNAi to deplete both ITSN1 isoforms (Fig. 5A). Loss of ITSN1 resulted in a significant decrease in the ability of extracellular virus to nucleate an actin tail (Fig. 5B,C). This reduction could be fully rescued by expression of RNAi-resistant ITSN-1L but not ITSN-1S, suggesting that the GEF activity of ITSN1 is required (Fig. 5D). Importantly, depletion of ITSN1 did not affect recruitment of AP-2 to viral particles (Fig. 5E). Taken together, our observations suggest that although ITSN1 is recruited at the time of clathrin accumulation, its role appears distinct from the clustering activity of AP-2 and clathrin in actin tail formation.
ITSN1 functions in the same pathway as Cdc42
To investigate whether ITSN1 promotes actin tail formation by activating Cdc42, we examined the consequence of expressing dominant-negative Cdc42-N17 in the absence of ITSN1 (Fig. 6A). Depletion of ITSN1 or expression of GFP–Cdc42-N17 resulted in an equivalent decrease in the number of extracellular viruses inducing actin tails. When combined, there was no enhanced reduction, suggesting that ITSN1 and Cdc42 function in the same pathway (Fig. 6A). To examine possible cooperative interactions between Cdc42, ITSN1 and N-WASP, we performed pull down assays on HeLa cells expressing GFP-tagged versions of N-WASP in cells depleted of endogenous N-WASP by RNAi (Fig. 6B). Cells were additionally expressing GST–Cdc42 to allow for examination of its interaction with N-WASP and the H208D mutant. As expected, we found that introduction of the H208D mutation into GFP–N-WASP eliminated binding to GST–Cdc42. In contrast, the H208D mutation had no impact on the interaction of GFP–N-WASP with endogenous ITSN1 (Fig. 6B). Depletion of ITSN1, however, diminished binding of GST–Cdc42 to GFP–N-WASP (Fig. 6B). These data are consistent with ITSN-1L acting as a GEF for Cdc42 to promote the interaction of its GTP-bound form with N-WASP (Hussain et al., 2001). Taken together, our data suggest that, once recruited by Nck, through WIP, N-WASP engages with ITSN1, activating Cdc42. This, in turn, binds and stabilises N-WASP, with all three components acting together in a feed-forward loop.
ITSN1 promotes N-WASP-mediated Fc gamma receptor phagocytosis
Given our observations with vaccinia, we decided to examine whether Cdc42, ITSN1 and Nck also act cooperatively during an N-WASP-dependent cellular process outside the context of infection. Recent work has shown that Nck and Cdc42 act synergistically to promote N-WASP-dependent phagocytosis mediated by the Fc gamma receptor (Dart et al., 2012). We therefore investigated whether ITSN1 also contributes to this pathway. We found that GFP–ITSN-1L is recruited to phagocytic cups along with GFP-tagged Nck, Cdc42 and N-WASP, as has been previously described (Fig. 7A) (Dart et al., 2012). To determine whether ITSN1 participates in phagocytic uptake, we depleted ITSN1 as above (Fig. 5A) and quantified the ability of HeLa cells transiently expressing the Fc gamma receptor IIa to phagocytose sheep red blood cells (RBCs) (Fig. 7B). Although siRNA treatment did not affect the attachment of RBCs to HeLa cells (Ai, the total number of RBCs per cell), it did lead to a decrease in the number that were internalised (Pi, number of internalised RBCs per cell) (Fig. 7B). Consequently, the percentage phagocytosis was significantly reduced, showing that ITSN1 enhances phagocytic uptake.
To determine whether ITSN1 regulates the same signalling pathway within phagocytosis as during vaccinia actin tail formation, we combined its depletion with that of Nck1, Cdc42 or N-WASP (Fig. 7C). As previously observed, depletion of Cdc42 led to a significant decrease in phagocytosis (Fig. 7D; supplementary material Fig. S3) (Dart et al., 2012). The combined loss of ITSN1 with Cdc42, however, led to no further reduction in phagocytosis (Fig. 7D; supplementary material Fig. S3). Furthermore, the same phenotype was observed when ITSN1 depletion was combined with loss of Nck or N-WASP (Fig. 7D; supplementary material Fig. S3). These data confirm that ITSN1 contributes to the actin polymerisation driven by Nck, Cdc42 and N-WASP during phagocytosis mediated by the Fc gamma receptor.
N-WASP, and the related haematopoietic-specific WASP, play important roles in stimulating Arp2/3-mediated actin polymerisation during a variety of cellular processes, such as endocytosis and phagocytosis, as well as during invadopodia and podosome formation (Mizutani et al., 2002; Benesch et al., 2005; Yamaguchi et al., 2005; Park and Cox, 2009). N-WASP is also essential for actin-based motility of a variety of intracellular pathogens (Lommel et al., 2001; Snapper et al., 2001; Lommel et al., 2004; Stamm et al., 2005; Dodding and Way, 2009). We have now shown that both vaccinia virus and the Fc gamma receptor use similar pathways to recruit and activate N-WASP during Arp2/3-mediated actin polymerisation. In both cases, Nck and Cdc42 act in a cooperative manner, with the activity of Cdc42 additionally regulated by the GEF ITSN1. Nck and Cdc42, however, appear to play different primary roles within the vaccinia-signalling network, even though both proteins are capable of activating N-WASP in vitro (Egile et al., 1999; Rohatgi et al., 1999; Rohatgi et al., 2001).
Nck is recruited to the virus in response to Src- and Abl-family-kinase-mediated phosphorylation of tyrosine 112 in the viral transmembrane protein A36 (Frischknecht et al., 1999; Newsome et al., 2004; Reeves et al., 2005; Newsome et al., 2006). Vaccinia is unable to induce actin tail formation in the absence of phosphorylation of tyrosine 112 or in cells lacking Nck, as the latter is essential to recruit WIP–N-WASP to facilitate activation of the Arp2/3 complex (Frischknecht et al., 1999; Ward and Moss, 2001a; Scaplehorn et al., 2002; Weisswange et al., 2009). Once recruited by Nck, the WIP–N-WASP complex is further stabilised by an interaction with Grb2, a downstream response to tyrosine 132 phosphorylation in A36 (Scaplehorn et al., 2002; Weisswange et al., 2009; Donnelly et al., 2013). In contrast to Nck, Grb2 is not essential for actin tail formation, although it does enhance the ability of the virus to induce actin polymerisation, consistent with its capacity to bind and activate N-WASP (Carlier et al., 2000).
Nck contains three SH3 domains that are capable of interacting with the proline-rich regions of WIP and N-WASP (Antón et al., 1998; Rohatgi et al., 2001; Donnelly et al., 2013). Our recent observations have, however, shown that the WIP–WASP complex is recruited to the virus by an interaction of the second SH3 domain of Nck with WIP (Donnelly et al., 2013). N-WASP is thus recruited to the virus because it is bound to WIP, rather than by directly binding to Nck in the first instance. Furthermore, the loss of the ability of Nck to bind N-WASP did not inhibit vaccinia actin tail formation, although it did reduce tail length and slightly increased the turnover rate of N-WASP on the virus (Donnelly et al., 2013). The turnover rate of an N-WASP mutant that is deficient in Nck binding, however, still did not approach that of Nck and WIP, even in the absence of Grb2 (Weisswange et al., 2009; Donnelly et al., 2013). These observations immediately suggested that an additional and more dominant factor is involved in stabilising and activating N-WASP.
Our observations examining the ability of N-WASP-H208D to induce actin tails in N-WASP−/− MEFs, as well as the decreased rate of vaccinia actin-based motility in this background, have now clearly demonstrated that Cdc42 is the missing factor. Consistent with this, loss of the ability of N-WASP to bind Cdc42 has a greater impact on virus actin tail formation than the loss of Nck binding, suggesting that Cdc42 plays a more important role than Nck in stabilising and activating N-WASP within the signalling network. Nevertheless, Nck still contributes, as loss of the ability of N-WASP to bind Cdc42 and Nck has a larger impact than that of Cdc42 alone. The combined loss of Cdc42 and Nck binding to N-WASP did not, however, fully abrogate actin polymerisation, even in the absence of Grb2 recruitment by vaccinia expressing A36-Y132F. This suggests that additional factor(s) besides Nck, Cdc42 and Grb2 can contribute to the ability of N-WASP to activate the Arp2/3 complex during vaccinia actin tail formation.
Previous in vitro observations have demonstrated that phosphatidylinositol (4,5)-bisphosphate [PtdIns(4,5)P2] can synergise with both Nck and Cdc42 to promote N-WASP activation (Rohatgi et al., 1999; Prehoda et al., 2000; Rohatgi et al., 2000; Rohatgi et al., 2001; Papayannopoulos et al., 2005; Tomasevic et al., 2007; Rivera et al., 2009). The ability of N-WASP to be activated by Cdc42, PtdIns(4,5)P2 and SH3-containing adapters, such as Nck and Grb2, allows it to connect a wide variety of signalling inputs to Arp2/3-mediated actin polymerisation. Moreover, the ability of multiple inputs to cooperatively activate N-WASP allows for exquisite regulation of Arp2/3-dependent actin polymerisation (Prehoda et al., 2000; Rohatgi et al., 2000; Papayannopoulos et al., 2005; Padrick et al., 2008; Padrick and Rosen, 2010). The relative contribution of these different inputs in activating N-WASP in vitro, however, varies depending on the individual study. More importantly, the majority of these studies have been performed in the absence of WIP and might therefore not fully reflect the situation in vivo. Indeed, in vitro, Cdc42 alone is not able to activate a WIP–N-WASP complex to stimulate Arp2/3 (Ho et al., 2004). It would be surprising, given previous studies, if PtdIns(4,5)P2 did not contribute to vaccinia actin tail formation by helping to stabilise N-WASP in an open active conformation. It should, however, not be forgotten that the SH3 domains in ITSN1, as well as those in Src and Abl family kinases, might also contribute activating inputs into N-WASP in the absence of Cdc42 and Nck. Indeed, the SH3 domain of the tyrosine kinase Arg, which is recruited by vaccinia (Reeves et al., 2005; Newsome et al., 2006), is capable of activating N-WASP in vitro (Miller et al., 2010). Nevertheless, our analysis suggests that in the absence of Grb2 recruitment, Arp2/3-dependent actin based motility of vaccinia is principally driven by Cdc42- and Nck-dependent activation of N-WASP.
We have recently observed that vaccinia transiently recruits AP-2 and clathrin prior to actin tail formation (Humphries et al., 2012). Given this, we examined whether the virus activates Cdc42 by recruiting ITSN1, a Cdc42 GEF that interacts with both AP-2 and N-WASP (Hussain et al., 2001; Hunter et al., 2013). We found that ITSN1 is recruited upstream of actin tail formation as it can associate with extracellular viruses attached to the plasma membrane in N-WASP−/− MEFs. The upstream localisation of ITSN1 is consistent with its role as an early endocytic protein, suggesting the same signals that are responsible for AP-2 and clathrin recruitment are also responsible for the initial recruitment of ITSN1 (Henne et al., 2010; Cocucci et al., 2012). However, ITSN1 was not required for the AP-2- and clathrin-dependent regulation of actin tail formation. This is probably because of the presence of other early endocytic proteins that could act in a redundant manner with ITSN1 to regulate AP-2 and/or clathrin recruitment and/or organisation.
ITSN1 is associated with the virus prior to actin tail formation. We therefore envisage that ITSN-1L promotes the activation of Cdc42 upon engagement with N-WASP, which is recruited as a complex with WIP downstream of Nck (Donnelly et al., 2013). This is consistent with a model put forward by Hussain et al. (Hussain et al., 2001), which suggests that the interaction of ITSN-1L with N-WASP enhances its GEF activity towards Cdc42. The local activation of Cdc42 in turn contributes to N-WASP activation, in addition to providing a stabilising interaction to maintain Arp2/3-mediated actin polymerisation. Such a scheme would account for the lag in Cdc42 recruitment to the virus during initiation of actin polymerisation as compared to that of Nck and N-WASP. A similar situation also appears to be the case for phagocytosis mediated by the Fc gamma receptor, as Cdc42 failed to robustly localise to phagocytic cups in the absence of N-WASP (Dart et al., 2012).
Studies on neuronal cells have shown that ITSN-1L can activate Cdc42 to promote secretory granule exocytosis in an N-WASP-dependent manner, and expression of a DH-domain-containing construct in neuroblastoma cells stimulates Cdc42-dependent neurite extension in response to serum starvation (Hussain et al., 2001; Gasman et al., 2004; Malacombe et al., 2006). Our observations, with vaccinia and phagocytosis mediated by the Fc gamma receptor, suggest that ITSN1 is likely to have a more general role in other Cdc42–N-WASP-dependent processes beyond neuronal cell lineages. Moreover, using vaccinia has allowed us to dissect the relative contribution of Cdc42 and Nck in promoting N-WASP-dependent actin polymerisation through the Arp2/3 complex. The conservation of this signalling network during phagocytosis mediated by the Fc gamma receptor further supports the use of vaccinia in elucidating mechanisms regulating N-WASP-dependent activation of the Arp2/3 complex.
MATERIALS AND METHODS
Antibodies against the following proteins were used for immunofluorescence and immunoblotting: Cdc42 (P-1, Santa Cruz Biotechnology, USA), B5 (Hiller and Weber, 1985), N-WASP (Moreau et al., 2000), ITSN1 (Hussain et al., 1999, and 118262, AbCam, UK), Cortactin (4F11, Millipore, USA), AP-2 (2730, AbCam, UK), Nck (06-288, Millipore, USA), GST (G7781, Sigma-Aldrich, USA), β-actin (AC-74, Sigma-Aldrich, USA), GFP (3E1, Cancer Research UK). Secondary antibodies were from Invitrogen, USA. Phalloidin-conjugated Texas Red (Invitrogen, USA) was also used.
Constructs and stable cell lines
The pE/L-GFP-Cdc42 and pE/L-GST-Cdc42 constructs have been previously described (Moreau et al., 2000; Arakawa et al., 2007a). Human intersectin-1L and intersectin-1S clones were provided by Stephane Gasman (INCI, France) (Hussain et al., 2001; Malacombe et al., 2006). GFP-tagged human ITSN-1L and ITSN-IS were generating by PCR by inserting the product into the NotI-SpeI sites of the vaccinia expression vector pE/L-GFP (Frischknecht et al., 1999). RNAi-resistant pE/L-GFP-ITSN-1L/-IS expression constructs and pLVX-GFP-N-WASP-H208D constructs were generated by site-directed mutagenesis using the Quikchange mutagenesis protocol (Stratagene, USA) (Donnelly et al., 2013). For RNAi-resistant constructs the sequence corresponding to the second ITSN1 siRNA oligonucleotide was subjected to mutagenesis at each base pair to alter the DNA sequence but not its protein coding. Stable GFP–N-WASP-expressing cell lines were established as previously described (Weisswange et al., 2009; Humphries et al., 2012).
Cells were infected with the WR strain of vaccinia virus, WR expressing A36-Y132F or with WR expressing an RFP-tagged version of the viral core protein A3 (Scaplehorn et al., 2002; Weisswange et al., 2009). HeLa cells and MEFs were fixed at 9 and 15 hours post-infection respectively. Cells were processed for immunofluorescence as previously described (Arakawa et al., 2007b; Arakawa et al., 2007a). For assays involving transfection, cells were transfected 4 hours post-infection with the indicated construct using effectene (Qiagen, Germany) and fixed 9 hours post-infection. siRNA treatment was as previously described (Humphries et al., 2012), briefly cells were transfected with 20 nM siRNA using the HiPerFect fast-forward protocol (Qiagen, Germany) and treated for 48 hours before use. The following siRNA oligonucleotides were used: ITSN1 (5′-GGCCAUAACUGUAGAGGAA-3′) and (5′-GAUAUCAGAUGUCGAUUGA-3′), AP-2 (5′-AGCAUGUGCACGCUGGCCA-3′) (Humphries et al., 2012). Pools of siRNA were obtained commercially (Dharmacon, USA). Plaque assays were performed as previously described, with the exception that stable MEF lines were used instead of A549 cells (Humphries et al., 2012). Live imaging and FRAP experiments were also performed as previously described (Humphries et al., 2012).
To analyse the interaction between N-WASP, ITSN1 and Cdc42, 6-cm dishes of HeLa cells were treated with N-WASP siRNA and either control or ITSN1 siRNA for 56 hours. Cells were then infected with WR virus, and 1 hour post infection transfected with pE/L expression vectors encoding GFP, GFP–N-WASP or GFP–N-WASP-H208D. Cells were additionally transfected with pE/L-GST-Cdc42. Cells were lysed and the GFP-trap experiment was performed according to the manufacturer's recommendation (Chromotek, Germany). The samples were then probed with antibodies against GFP and ITSN, and GST to look at the interaction with overexpressed GST–Cdc42.
Cells were transfected with the Fc gamma receptor IIa (Dart et al., 2012), using fugene 6 (Promega, USA). At 20 hours after transfection, the phagocytosis assay was performed. Sheep RBCs (TCS Biosciences, UK) were washed with gelatin veronal buffer (Sigma-Aldrich, USA) and incubated with rabbit anti-RBC IgG (MP Biomedicals, France) on a rotating wheel for 30 minutes. Meanwhile, HeLa cells were serum starved for 1 hour. RBCs were washed again, resuspended in serum-free medium, and then added to the cells. Cells were incubated at 4°C for 15 minutes, then at 37°C for 30 minutes before fixation. The internalisation of RBCs was determined by staining with a secondary antibody pre- and post-permeabilisation. The data are normalised with respect to control; Ai represents the total number of RBCs per cell, and Pi the number of internalised RBCs per cell. The percentage phagocytosis is the number of internalised RBCs as a percentage of total RBCs per cell.
Microscope image acquisition
For imaging of fixed samples, a Zeiss Axioplan2 microscope equipped with a Photometrics Cool Snap HQ cooled CCD camera, external Prior Scientific filter wheels (DAPI; FITC; Texas Red; Cy5) and a 63× 1.4 NA Plan Achromat objective was used. The system was controlled with MetaMorph 6.3r7 software. Live-cell imaging and FRAP was carried out on a Zeiss Axio Observer microscope equipped with a Plan Achromat 63× 1.4 NA Ph3 M27 oil lens, an Evolve 512 camera and a Yokagawa CSUX spinning disk. Imaging was carried out at 37°C, in Phenol-Red-free MEM with 40 mM HEPES. The system was controlled by Slidebook 5.0, movies were analysed using either the Slidebook or MetaMorph software.
Quantification and figure preparation
Cells were analysed on the basis of an established viral factory using DAPI staining, to allow determination of infection. The percentage of cells inducing actin tails, and the number of tails was determined manually. For the percentage of cells, 100 cells were counted in three separate experiments. For the number of tails and percentage of extracellular virus inducing actin, ten cells each were counted in three separate experiments. To measure the intensity of GFP-tagged Cdc42, Nck and N-WASP during actin tail formation, the intensity was recorded for 30 different nucleation events using Slidebook software (3i Intelligent Imaging Innovations, USA). The GFP signal was normalised to the maximum intensity. Graphs and statistics were compiled using Prism (Graphpad software, USA), comparison of two data sets was carried out using a Student's t-test and, for larger data sets, a one-way ANOVA was performed with multiple comparison analysis. FRAP analysis was as previously described (Weisswange et al., 2009; Humphries et al., 2012). Plaque size was measured using ImageJ (NIH, USA). All figures were generated using Adobe software (Adobe, USA). P<0.05, P<0.01 and P<0.001 is represented by single, double and triple asterisks, respectively.
The authors would like to thank Anna Dart (King's College London, UK) for help in establishing the phagocytic assay and for the Fc gamma receptor IIa construct. We would also like to thank Scott Snapper (MGH, USA) for N-WASP−/− cell lines, Stephane Gasman (INCI, France) for ITSN1 clones and Peter McPherson (MNI, McGill University, Canada) for the ITSN1 antibody. In addition, we would like to thank members of the Way laboratory for constructive comments on the manuscript.
A.C.H., S.K.D. and M.W. designed experiments. A.C.H. and S.K.D. performed the experiments and data analysis. A.C.H. and M.W. wrote the paper.
This work was supported by Cancer Research UK and LRI Ph.D. studentships to A.C.H. and S.K.D.
The authors declare no competing interests.