Salmonella invasion of a cell is self-limiting due to effector-driven activation of N-WASP

Summary Salmonella Typhimurium drives uptake into non-phagocytic host cells by injecting effector proteins that reorganize the actin cytoskeleton. The host actin regulator N-WASP has been implicated in bacterial entry, but its precise role is not clear. We demonstrate that Cdc42-dependent N-WASP activation, instigated by the Cdc42-activating effector SopE2, strongly impedes Salmonella uptake into host cells. This inhibitory pathway is predominant later in invasion, with the ubiquitin ligase activity of the effector SopA specifically interfering with negative Cdc42-N-WASP signaling at early stages. The cell therefore transitions from being susceptible to invasion, into a state almost completely recalcitrant to bacterial uptake, providing a mechanism to limit the number of internalized Salmonella. Our work raises the possibility that Cdc42-N-WASP, known to be activated by numerous bacterial and viral species during infection and commonly assumed to promote pathogen uptake, is used to limit the entry of multiple pathogens.


INTRODUCTION
Salmonella enterica is a facultative intracellular bacterial pathogen that causes a range of diseases in both humans and animals. 1 S. enterica serovar Typhimurium (S. Tm) typically causes acute inflammatory gastroenteritis but can also induce a more serious systemic disease. 2 Once ingested through e.g. contaminated food, S. Tm takes advantage of the host's inflammatory response to gain a competitive advantage over the resident microbiota in the gut lumen. 3,4 This inflammatory response is driven by the invasion of non-phagocytic intestinal epithelial cells by a fraction of the infecting pathogen population. 5,6 S. Tm has a multitude of likely redundant mechanisms to stimulate actin-dependent uptake into cells. 7 The best-characterized mode of S. Tm entry into epithelial cells uses a type 3 secretion system to inject a myriad of effector proteins that manipulate and mimic host proteins involved in the regulation of the actin cytoskeleton. 8 For example, the delivered effectors SipA and SipC directly bind actin, cooperatively nucleate actin polymerization and bundle actin filaments. 9 In contrast, SopE1 and SopB manipulate the cytoskeleton indirectly 10 -SopE1 is a guanine nucleotide exchange factor (GEF), activating host GTPases such as Rac1, 11 while SopB has lipid phosphatase/phosphotransferase activity 12,13 that results in the recruitment and activation of the GTPase Arf1. 14 Rac1 and Arf1 cooperate to activate the WAVE regulatory complex (WRC), 14,15 one of the cell's major nucleation promoting factors (NPFs), that stimulate the activity of the branched-chain actin nucleator the Arp2/3 complex. The resulting generation of distinct actin-rich membrane protrusions 16 encapsulate and internalize the bacteria in a manner similar to macropinocytosis. 17 A second NPF, WASH, has also been shown to play a role in the actin rearrangements underlying cell invasion, though the precise mechanism by which S. Tm manipulates this pathway remains uncertain. 18  (F) Uptake of S. Tm into WT MEFs ectopically expressing a control empty vector (Empty), Cdc42 QL or Rac1 QL . Cells were pretreated with PBS (Ctrl) or 10 mM 187-1 (+187-1). All values relative to MEFs transfected with a control empty vector. All S. Tm uptake was measured 30 min after initial infection. Values are the means of three independent replicates (each replicate is comprised of 6 fields of view, n = 18. Each field of view is approximately 50-100 cells, see method details). Error bars indicate standard deviation. NS -no significant difference, ** -p < 0.01 (ANOVA followed by post hoc Dunnett's comparison). ll OPEN ACCESS 2 iScience 26, 106643, May 19, 2023 iScience Article 187-1 led to a comparable increase in S. Tm entry to that seen for Hap1 and MEF cells ( Figure 1D). As expected, the addition of 187-1 to DN-WASP MEF or Hap1 cells had no additional effect, however 187-1 treatment of both DAbi1 MEFs and DNap1 Hap1 cells led to a significant increase in S. Tm cell entry ( Figure S1C). Therefore, it appears that N-WASP inhibition actually enhances both WRC-dependent and -independent uptake of S. Tm.
Consistent with this inhibitory role, ectopic expression of wild-type (WT) N-WASP in either WT MEF or Hap1 cells ( Figure S1D) impeded S. Tm uptake by approximately 40%, and the expression of a constitutively active N-WASP mutant (N-WASP L299P ) 26 caused an even greater defect ( Figure 1E). Expression of the Arp2/3 complex-binding VCA domain of N-WASP (VCA) caused a 20% inhibition, but when fused to a phosphoinositide-binding pleckstrin homology (PH) domain (that localises the VCA to the plasma membrane-PH-VCA, see method details) uptake was more severely impaired ( Figure 1E). Likewise, expression of a constitutively active form of the N-WASP activator Cdc42 (Cdc42 QL ) impeded S. Tm uptake by approximately 70%, an effect completely reversed in the presence of the N-WASP inhibitor 187-1 ( Figure 1F). Conversely, the expression of control WRC-activating Rac1 QL had no significant effect on S. Tm uptake.
Collectively these results demonstrate that N-WASP, activated by Cdc42, is inhibitory for S. Tm entry into mammalian cells.

N-WASP-dependent inhibition of Salmonella uptake is driven by SopE2
The inhibitory role of Cdc42-N-WASP is somewhat surprising as S. Tm delivers several effector proteins that can activate Cdc42. 27 The S. Tm strain SL1344 used in this study expresses two guanine nucleotide exchange factors (GEFs): SopE1, that activates both Rac1 and Cdc42, and SopE2 that activates only Cdc42. 27 To determine the contribution of these GEFs to Cdc42 activation during cell entry, WT MEFs were infected with SL1344 strains lacking either sopE1, sopE2 or both. A GTPase activation assay determined that whilst WT or DsopE1 strains effectively activated Cdc42 in the cell, DsopE2 and DsopE1/E2 strains did not, indicating that SopE2 is largely responsible for S. Tm driven activation of Cdc42 ( Figure 2A).
To determine whether this SopE2-dependent activation of Cdc42 resulted in increased N-WASP activation, GFP-Cdc42 was isolated from uninfected and infected cells using a GFP Trap (Chromotek) and the level of bound N-WASP measured by immunoblotting. Infection of cells with WT but not DsopE2 S. Tm resulted in a substantial increase in N-WASP binding to Cdc42 ( Figure 2B), indicative of SopE2-dependent N-WASP activation. Consistent with this, ectopic expression of SopE2, but not SopE1, (Figure S2A), caused a 50% decrease in S. Tm uptake, which was completely reversed in the presence of the N-WASP inhibitor 187-1 ( Figure 2C).
As exogenous SopE2 expression leads to N-WASP activation, we next explored how native SopE2 delivered by S. Tm contributes to entry. As previously demonstrated, 11 the control deletion of sopE1 (DsopE1) impeded invasion into WT MEFs by over 55%, yet the loss of sopE2 (DsopE2) promoted invasion by over 25% ( Figure 2D). Similar results were also observed in Hap1 cells ( Figure S2B). The increase in invasion caused by the loss of SopE2 expression was not observed in DN-WASP MEFs, or WT MEFs treated with 187-1, strongly suggesting that the inability of S. Tm DsopE2 to activate N-WASP is responsible for the increased invasion ( Figure 2E). Collectively, these results indicate that activation of N-WASP by delivered SopE2 is inhibitory for S. Tm entry.

SopE2 and N-WASP inhibit Salmonella uptake at later timepoints of invasion
The finding that the ''entry'' effector SopE2 is antagonistic to S. Tm invasion is somewhat unexpected. To gain a deeper insight into the role of SopE2, S. Tm invasion assays were performed using infection lengths ranging from 5 to 120 min with both WT and DsopE2 strains ( Figure 3A). After a 5-or 15-min infection there was no significant difference in invasion by WT or DsopE2 S. Tm, yet after 30 min there was a clear, approximately 20% increase in the entry of DsopE2 as we have already demonstrated ( Figure 2D). Strikingly, the uptake of WT S.  We hypothesized that in our assay, after 30-60 min S. Tm-delivered SopE2 is able to activate sufficient N-WASP to block further bacterial uptake from occurring. Interestingly, the characteristic membrane ruffles formed by S. Tm to drive invasion failed to form in WT MEFs that had been pre-infected for 1 h with unstained WT S. Tm. Ruffles did however form in DN-WASP MEFs that had been pre-infected, as well as in WT and DN-WASP MEFs that had been pretreated with PBS as a control ( Figure 3C). In control PBSpretreated WT and DN-WASP MEFs, and DN-WASP MEFs pre-infected with WT S. Tm, approximately 60-75% of fresh S. Tm were associated with actin-rich ruffles, with this number falling to 15% in WT MEFs pre-infected with WT S. Tm ( Figures 3D, S3C). To demonstrate that the inability to form membrane ruffles in pre-infected cells also affected S. Tm uptake, double infection invasion assays were carried out (illustrated in Figure 3E). Briefly, WT or DN-WASP MEFs were pretreated with PBS, or pre-infected with WT or DsopE2 S. Tm for 60 min, before being treated with gentamycin for 30 min to kill extracellular bacteria. Subsequently, these pre-infected cells were subjected to a second 30-min infection with fresh WT S. Tm that express GFP only when internalized, 28 allowing the quantitation of their invasion. Pre-infection with WT S. Tm reduced subsequent invasion by fresh bacteria by almost 90%, compared to control cells, while cells that were pre-infected with DsopE2 S. Tm, were invaded similarly to control cells ( Figure 3F). The ability to seemingly make the cell impervious to further rounds of S. Tm uptake lasts for at least 4 h after the initial infection ( Figure S3D), and importantly, pre-infection of DN-WASP cells by either WT or DsopE2 S. Tm had no effect on invasion of fresh WT S. Tm ( Figure 3F). An inhibition in the uptake of WT S. Tm was also clear in WT MEFs, but not DN-WASP MEFs that were preinfected with Enteropathogenic Escherichia coli (EPEC), a pathogen well known to activate N-WASP for actin pedestal formation required for cellular attachment 29 ( Figure S3E).
Collectively, these data suggest that N-WASP-dependent actin assembly induced by SopE2 delivery makes cells resistant to further pathogen uptake. To test this hypothesis, we attempted to restore invasion susceptibility to pre-infected cells. To achieve this, cells pre-infected with WT S. Tm (as above) were treated for 15 min with either control DMSO, the N-WASP inhibitor 187-1 or the actin destabilizing drugs Cytochalasin D 30 or Latrunculin A. 31 The cells were then thoroughly washed to remove the drugs, and then infected with fresh WT S. Tm. Strikingly, the inhibition of S. Tm invasion observed in pre-infected cells was almost completely reversed when cells were treated with Cytochalasin D, Latrunculin A or 187-1 but not by a DMSO control ( Figure 3G). This strongly implies that actin assembled by N-WASP in response to infection impedes further S. Tm from stimulating their own actin-dependent uptake.

SopA controls how SopE2 affects Salmonella uptake
Our results demonstrate that SopE2 and N-WASP negatively regulate S. Tm uptake, but it is also apparent that this inhibition only occurs at later timepoints (30 min or later in our assay). SopE2 may therefore also play a role in promoting invasion at earlier timepoints. Consistent with this, deletion of sopE2 in a DsopE1 background results in a small additional defect in invasion after 15 min ( Figure 4A). Indeed, the DsopE1/E2 strain invades poorly after 5, 15 and 30 min, but at later timepoints, in the absence of the inhibitory SopE2-Cdc42-N-WASP pathway, internalization of S. Tm continues to occur, eventually overtaking that seen for WT S. Tm ( Figure S4A). This indicates that at early stages of infection SopE2 plays a small but positive role in invasion, an effect likely masked in strains where invasion is driven by the dominant SopE1. Cdc42 activated by SopE2 may promote this early invasion via N-WASP, or via other proteins activated by Cdc42 that drive actin assembly -such as formins (known to play a role in S. Tm uptake 32 ). To test this hypothesis MEFs treated with 187-1 or the pan-formin inhibitor SMIFH2 33 were infected with WT, DsopE1, and DsopE1/E2 S. Tm ( Figure 4A). N-WASP inhibition promoted invasion by all three S. Tm strains at 15 min. Interestingly however, both the DsopE1 and DsopE1/E2 strains invade to a similar extent in the presence of SMIFH2, indicating that in the absence of SopE1 at least, SopE2 activation of Cdc42 can promote invasion in a formin-dependent manner at early stages of an infection.
SopE2 therefore may play both a positive and negative role in S. Tm cell invasion. SopE2 is found in almost all serovars of S. enterica, 34 and its sequence is highly conserved ( Figure S4B), yet in serovar Typhi (S. Typhi) it is pseudogenised. 35 Consistent with our data here, the introduction of SopE2 from S. Tm into S. Typhi greatly impedes the uptake of S. Typhi into cultured cells. 35 More intriguingly this inhibition could be overcome through the expression of S. Tm SopA, a second effector which is also pseudogenised in S. Typhi. 35 This suggests that SopA may be able to prevent SopE2 from inhibiting S. Tm uptake and thus function to control the balance of positive and negative contributions of SopE2 to invasion.
To investigate this, we performed 5-120-min infection time-course experiments. The loss of SopA (DsopA) caused a modest, statistically significant, inhibition of invasion (15-35%) at all timepoints measured ( Figure 4B), consistent with previous reports. 36 The DsopE1 strain invades more poorly than WT at all iScience Article timepoints (55% less at 30 min) and like WT invasion does not continue after 60 min ( Figure 4C). The additional deletion of sopA (DsopE1DsopA) results in a further drop in invasion at all timepoints relative to DsopE1 ( Figure 4C, the two strains plotted alone for clarity in Figure S4C). However, the deletion of sopA in the strain lacking sopE2 (DsopE2DsopA) did not have a negative effect on invasion ( Figure 4C), indicating that SopE2 is responsible for the entry defect observed in the other strains lacking SopA. The invasion of DsopE1DsopA after 15 min is almost identical to that of DsopE1/E2 ( Figure S4D) which suggests that the positive contribution of SopE2 to uptake at early stages of invasion ( Figure 4A) is dependent on the presence of SopA. Therefore, in the absence of SopA, Cdc42-N-WASP signaling, driven by SopE2, is unimpeded, causing invasion to be inhibited, and any positive role for SopE2 at early stages is lost.
Consistent with the idea that SopA prevents Cdc42-N-WASP signaling from inhibiting invasion, WT S. Tm invasion of WT MEFs ectopically expressing SopA ( Figure S4E) continued to occur steadily 30-120 min after iScience Article the start of infection ( Figure 4D), as seen above for DsopE2 entry ( Figure 3A). Ectopic expression of SopA also resulted in the continuous invasion of DsopA S. Tm, as well as the continued uptake of the poorly invasive DsopE1. SopA expression did not have an additional effect on the already continuously invasive DsopE2 strain ( Figure S4F). SopA is a ubiquitin ligase, 37 and expression of an inactive variant (SopA C753A ) did not affect the uptake of WT S. Tm ( Figure 4D), suggesting that ubiquitination likely regulates SopE2 signaling.
Enteropathogenic. E. coli (EPEC) form actin pedestals and tightly adhere to target cells in an N-WASPdependent manner. 29 In cells expressing SopA, but not SopA C753A , EPEC pedestal formation was greatly compromised ( Figure 4E), with extremely small or completely absent pedestals. These effects were mirrored by the overall ability of EPEC to attach to cells, with attachment to cells expressing SopA being similar to that to DN-WASP cells (reduced by 83% and 88% respectively; Figure 4F, with representative images in Figure S4G). This data rules out the possibility that SopA interferes directly with SopE2, and instead, SopA must modulate the cellular Cdc42-N-WASP pathway, either directly or indirectly.
Together the data presented here demonstrate that the Cdc42-N-WASP pathway, activated by SopE2, is able to block S. Tm uptake into host cells. During early periods of invasion SopA, in a ubiquitin ligasedependent manner, is able to prevent this by interfering with Cdc42-N-WASP signaling, thus allowing S. Tm to efficiently invade cells.

DISCUSSION
The data here allow us to propose a model for how the host cell's susceptibility to invasion by S. Tm is determined by the Cdc42-N-WASP pathway ( Figure 5). We postulate that early during invasion, N-WASP activity is low, and the host cell is susceptible to efficient S. Tm uptake. We observe that, on average, 2-3 bacteria enter per cell during this phase in our assay conditions. At later timepoints, as N-WASP activity increases the cell becomes recalcitrant to further S. Tm uptake, and invasion is effectively blocked. This inhibition of S. Tm uptake is driven by the effector SopE2, which activates the Cdc42-N-WASP pathway. At early timepoints, SopE2-dependent activation of Cdc42 does not inhibit S. Tm uptake and instead, Cdc42 plays a minor role in promoting uptake, potentially via activation of formins ( Figure 4). The transition of a host  iScience Article cell from being susceptible to S. Tm invasion to a state that is almost totally recalcitrant to uptake is controlled by the activity of the S. Tm effector SopA. At early timepoints, SopA activity is high, and N-WASP activity is consequently low, despite the activation of Cdc42 by SopE2, meaning the host cell can be invaded. As the activity of SopA is diminished, and N-WASP activity increases due to the action of SopE2, a threshold is met at which point the cell becomes impervious to further S. Tm uptake. In other words, although SopE2 activates Cdc42 at all phases of entry, this only triggers N-WASP-dependent actin assembly (and consequent inhibition of further uptake) at later timepoints, i.e. when SopA activity is reduced.
We demonstrate throughout this work that the activation of N-WASP, and subsequent N-WASP driven actin assembly, is inhibitory to S. Tm-driven uptake into host cells. It is likely that the precise level of Cdc42/N-WASP activity required to prevent pathogen uptake, and the time taken to achieve this level during infection will vary between different cell types and/or culture conditions. As N-WASP triggers actin assembly and is recruited to S. Tm entry foci, 20 it is generally assumed that N-WASP, and its activation by SopE2, positively contributes to invasion. 20,38,39 The only direct evidence for this is one study showing that invasion is inhibited by expression of a ''dominant-negative'' N-WASP construct, however this construct non-productively binds Cdc42, and thus would inhibit all Cdc42-dependent signaling, not just that to N-WASP. 20 More recently, and consistent with our results, it has been reported that S. Tm entry into N-WASP knockout cells is actually enhanced compared to wild-type cells. 18 Determining precisely why N-WASP-driven actin assembly is inhibitory to S. Tm forced uptake will require further study. It is logical that the activation of N-WASP and generation of actin filaments would result in a competition for resources, such as the Arp2/3 complex, capping proteins, cofilin, and free actin. Additionally, N-WASP is well-known to play a role in clathrin-mediated endocytosis 40 and could potentially recycle important actin machinery away from the plasma membrane. The actin generated by N-WASP activity may also act as a physical barrier that blocks other forms of actin assembly, or indeed the efficient delivery of effectors from extracellular S. Tm. Indeed, cortical actin assembly increases cell rigidity, which in turn can restrict the uptake of S. Tm into macrophages, 41 and Pseudomonas aeruginosa into lung epithelial cells. 42 It has also been reported that WASP, the Dictyostelium N-WASP homolog, is able to restrict Rac1 activity, 43 and it is well established that Rac1 is a key driver of S. Tm uptake. 44 Many other invasive bacteria and viruses activate N-WASP during the course of their own infection of host cells 45 and it will be intriguing to explore the role of N-WASP in these systems.
Unlike its homolog SopE1, whose role in promoting S. Tm uptake is well-established, the contribution of SopE2 to entry is less clear. Deletion of sopE2 in strains that express SopE1, e.g. SL1344 used here, has little effect on S. Tm entry. 22 Likewise, Cdc42, the target of SopE2's GEF activity, is not required for S. Tm uptake. 44,46 In the absence of sopE1, deletion of sopE2 has a more pronounced negative impact, 34 and in strains engineered to lack multiple effector proteins, expression of SopE2 alone is able to restore invasion of various cultured cells to approximately 10-20% of WT. 38 In contrast, the introduction of S. Tm sopE2 into S. Typhi, in which sopE2 is pseudogenised, results in reduced uptake of the bacteria. 35 Recent work has demonstrated that, in a strain lacking SopE1, both SopE2 and the actin-binding effector SipA play key, but redundant, roles in S. Tm invasion of the intestinal epithelium of neonate mice. 47 However, in adult mice invasion (by a sopE1 positive strain) can be severely reduced by the deletion of sipA alone, suggesting that SopE2 plays a less significant role. 48 Consistent with all these findings, our data show that SopE2 activation of Cdc42 has a small, N-WASP-independent, contribution to invasion at early timepoints, and that this is more evident in strains lacking sopE1 ( Figure S4). We propose therefore that Cdc42 activation by SopE2 initially promotes S. Tm uptake to varying degrees in a strain-, cell type-and/or developmental stage-dependent manner. Our data suggest this may be due to the activation of formins (Figure 4), but it could also potentially involve further downstream substrates of Cdc42, such as p21-activated kinase. 49 Eventually however, the activation of Cdc42-N-WASP by SopE2 reaches a threshold that leads to the cell becoming recalcitrant to further bacterial uptake.
While we can't rule out a role for changes in SopE2 level and/or localization, 50 it seems likely that the transition of Cdc42 from promoter to inhibitor of invasion is controlled by SopA as in the absence of SopA the small positive contribution of SopE2-Cdc42 at early stages of invasion isn't evident ( Figure S4C), whereas in cells ectopically over-expressing SopA the N-WASP-dependent inhibition of uptake is prevented (Figure 4). This suggests that the transition to the dominant inhibitory N-WASP activity at later timepoints of WT iScience Article infection could be linked to a loss of SopA activity in the cell. SopA is a ubiquitin ligase that can both autoubiquitinate and be a substrate for the host ubiquitin ligase RMA1 51,52 both of which may contribute to degradative ubiquitination which has been observed during infection. 53 In addition, the small RNA IsrM can repress SopA expression, 54 and is responsible for turning off SopA production at later stages of S. Tm infection of murine macrophages. 55 SopA levels are thus tightly controlled both pre-and post-translocation into host cells. It is also possible that the changes in SopA activity seen at later timepoints are due to altered SopA localization. The precise dynamics of SopA activity in the cell and how this correlates with the invasion potential of cells therefore requires further study, as does the precise mechanism by which SopA regulates signaling downstream of SopE2. This requires its ubiquitin ligase activity ( Figure 4D), and as SopA can also inhibit EPEC attachment (Figures 4F and 4G) its target is not SopE2 itself but a downstream cellular factor such as Cdc42 or N-WASP. The most prominent host substrates for SopA identified from proteomic studies are themselves ubiquitin ligases, namely TRIM56 and TRIM65, and targeting of these allows S. Tm to modulate the host inflammatory response. 56,57 Neither TRIM56 nor TRIM65 have any known link to Cdc42-N-WASP signaling or any other cytoskeletal pathway, so are likely not responsible for the effects described here. In the specific cell line and conditions assessed, many other proteins showed low-level SopA-dependent ubiquitination, including Cdc42 and N-WASP. 57 Intriguingly, a global study of host proteins ubiquitinated in response to S. Tm infection showed that Cdc42 is highly ubiquitinated at early timepoints (30 min), though the ubiquitin ligase responsible was not identified. 53 Irrespective of the precise mechanism, the idea that an intracellular pathogen such as S. Tm would restrict its own uptake is at first glance surprising. However, for many viruses, including HIV, 58 hepatitis C 59 and influenza 60 once a cell is infected, further infection is inhibited, a phenomenon termed superinfection exclusion. A similar mechanism has been described for the intracellular bacterial pathogen Chlamydia 61 and presumably may ensure maximal resources for replication for the initially invading pathogens. Interestingly, in models of systemic infection in mice, sites of S. Tm colonization such as the spleen and gall bladder generally only contain one or two bacteria per cell. 62,63 In humans S. Tm usually causes gastroenteritis and doesn't spread systemically. 3 S. Tm gastroenteritis is characterized by a localized inflammatory response triggered by the pathogen invading cells of the intestinal epithelium. However, the majority of the colonizing S. Tm remain in the lumen of the intestine where they take advantage of the inflammatory response to outcompete resident microbiota and spread to new hosts via the faeco-oral route. 5,6 Suppressing entry into already-infected cells may result in a greater proportion of infected cells, thereby amplifying the triggered immune response. In this scenario, it also makes sense that once a sufficient inflammatory response has been triggered, S. Tm has a mechanism to limit further unnecessary invasion, especially as the subpopulation that invade host cells are eventually killed, presumably by host immune cells such as neutrophils. In contrast to S. Tm, the disease caused by the human-adapted S. typhi (i.e. typhoid fever) is non-inflammatory-following entry into intestinal epithelium cells, the pathogen spreads systemically to colonize sites such as the gall bladder, from where the bacteria can be secreted into intestinal lumen and subsequently spread to new hosts. 64 Provocatively, both sopE2 and sopA have been pseudogenized in S. typhi, 35 suggesting that following the evolution of a systemic lifestyle this pathogen no longer requires a mechanism to limit cell invasion.
In conclusion, we show that rather than driving the entry of Salmonella into non-phagocytic cells, the Cdc42-N-WASP pathway, activated by the delivered effector SopE2, is instead inhibitory to pathogen uptake. N-WASP activation has been implicated in the invasion of host cells by many intracellular bacterial pathogens, 45 though as is the case for S. Tm, the direct evidence for this is often lacking. In fact, in at least one of these cases, namely Shigella flexneri, it has also been reported that exogenous expression of N-WASP is inhibitory to pathogen entry, 26 as we describe here for Salmonella. This raises the possibility that the specific actin assembly triggered by N-WASP causes host cells to become resistant to invasion by multiple bacterial pathogens. The manipulation of N-WASP signaling by various bacterial species, and how this contributes to pathogenesis, must therefore be carefully re-assessed.

Limitations of the study
We have shown that the effector-driven activation of N-WASP during S. Tm invasion of cells prevents the entry of further bacteria. We do not yet know how N-WASP is inhibitory, but this inhibition is apparent in multiple different cell types. Our data suggest that SopA supresses the inhibitory N-WASP pathway to allow S. Tm to enter cells at early timepoints, and we hypothesize that SopA activity is later diminished. However, as our antibodies are unable to detect the low levels of SopA delivered during infection we do not know if this is due to degradation or altered expression or localization. We cannot therefore model how this activity of SopA iScience Article coordinates with its other reported function in modulating the inflammatory response. Cell culture assays allow easy genetic manipulation of target cells, they also allow the study of cell entry with high temporal resolution and in isolation from factors such as the host immune system and competing microbiota. This is especially important when studying multifunctional effectors such as SopE2 and SopA which also play a role in modulating the host immune response to infection. However, understanding the importance of N-WASP-driven entry restriction, and its modulation by SopA, to disease pathogenesis will ultimately require in vivo studies.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:   iScience Article and pHA-SopA, were generated by amplification of the relevant gene from Salmonella SL1344 chromosomal DNA and cloning them into HA or Emerald-GFP vectors using Gateway methodology. HA-SopA C573A was generated using SDM of HA-SopA and confirmed via DNA sequencing.

Salmonella infection assays
Invasion was quantified using S. Tm carrying the pM975 plasmid. Once internalised these bacteria express GFP via a SPI2 promoter allowing for specific detection of internalised bacteria only. 28 Unless stated otherwise, cells were infected for 30 minutes at a multiplicity of infection (MOI) of 100, cells were washed twice in phosphate buffered saline (PBS), then incubated in media containing gentamycin (100 mg/ml) for 90 minutes. Cells were subsequently fixed, stained with Texas-Red Phalloidin (Life-Tech) and DAPI, then visualised using wide-field fluorescence microscopy with an Olympus IX81 inverted microscope. Using a 20x objective six fields of view per experimental repeat were utilised, with each field of view containing approximately 50-100 cells. The number of GFP-positive (internalised) bacteria and total number of cells (as determined from DAPI staining) per field of view were counted and the average number of bacteria per cell determined. All experiments were carried out a minimum of three times, the means calculated, and significance determined, with a P value of <0.05 (determined by ANOVA followed by a post hoc Dunnett's comparison) deemed significant.
Double infections were carried out by first infecting cells with S. Tm that did not harbour the pM975 plasmid for 60 minutes (or PBS as a control). Cells were then washed in PBS and incubated in media containing gentamycin for 30 minutes, washed again and media replaced. These cells were subsequently infected for 30 minutes with S. Tm carrying pM975 and an invasion assay carried out and quantified as described above. For drug washout experiments, cells were incubated for 15 minutes in media containing the indicated inhibitor after the first 60-minute infection. Cells were then washed three times in PBS, media was replaced, and the second infection carried out as above.
To investigate ruffle formation, cells were infected for 5 minutes with S. Tm that had been pre-stained with Alexa-Fluor-350 carboxylic acid succinimidyl ester (15 minutes), then washed in Tris-buffered saline (pH 7.4). Cells were stained with phalloidin to visualise actin, then analysed by wide-field fluorescence microscopy.

EPEC attachment assays
MEFs were infected with WT EPEC for 90 minutes, cells were then washed twice in PBS, twice in ice cold 200 mM glycine (pH 2), then again in PBS twice. Cells were fixed, stained with Texas-Red Phalloidin and DAPI, and adherent EPEC stained using an anti-intimin antibody. The number of adherent EPEC per cell were then counted using fluorescence microscopy (of at least 500 cells). All experiments were carried out at least three times, means calculated, and significance determined by ANOVA followed by a post hoc Dunnett's comparison (P <0.05 deemed significant).

GFP trap
MEFs were transfected to express GFP-tagged constructs 48 hours prior to performing the GFP trap, which was carried out according to the manufacturer's instructions (Chromotek). The resulting trapped proteins were analysed by SDS-PAGE before being transferred to PVDF membranes and probed with appropriate antibodies. All immunoblots were visualised using a LI-COR Odyssey Fc imaging system, utilising appropriate fluorescent secondary antibodies (LI-COR).

GTPase activation assay
Indicated cells were washed and lysed using RIPA buffer supplemented protease inhibitors . Cell lysates were clarified by centrifugation (13,000 g, 5 minutes) then incubated with GST-PBD (that specifically and strongly interacts with GTP-bound Rac1 and Cdc42) bound to glutathione-Sepharose resin at 4 C for 30 minutes. The resin was extensively washed with PBS and recruited proteins eluted with SDS-PAGE loading buffer supplemented with 6 M urea. Eluted (and therefore active) Cdc42 was detected using SDS-PAGE and immunoblotting.

ll
OPEN ACCESS