Shigella IpaH0722 E3 Ubiquitin Ligase Effector Targets TRAF2 to Inhibit PKC–NF-κB Activity in Invaded Epithelial Cells

NF-κB plays a central role in modulating innate immune responses to bacterial infections. Therefore, many bacterial pathogens deploy multiple mechanisms to counteract NF-κB activation. The invasion of and subsequent replication of Shigella within epithelial cells is recognized by various pathogen recognition receptors as pathogen-associated molecular patterns. These receptors trigger innate defense mechanisms via the activation of the NF-κB signaling pathway. Here, we show the inhibition of the NF-κB activation by the delivery of the IpaH E3 ubiquitin ligase family member IpaH0722 using Shigella's type III secretion system. IpaH0722 dampens the acute inflammatory response by preferentially inhibiting the PKC-mediated activation of NF-κB by ubiquitinating TRAF2, a molecule downstream of PKC, and by promoting its proteasome-dependent degradation.


Introduction
The intestinal epithelium deploys multiple sensors and defense systems against microbial intrusion. Microbial components and infection-associated cellular damage are recognized as pathogenassociated molecular patterns (PAMPs) and as danger-associated molecular patterns (DAMPs), respectively. Pattern recognition receptors (PRRs) recognize PAMPs and DAMPs, thus activating the immune system to clear the bacteria and initiate the repair of the injured epithelial lining [1,2]. Nevertheless, many bacterial pathogens such as Shigella, Salmonella, Yersinia, enteropathogenic Escherichia coli (EPEC), and enterohemorrhagic E. coli (EHEC), are able to efficiently colonize the intestinal epithelium by utilizing highly evolved mechanisms to counteract host innate defense mechanisms [3]. Previous studies have reported that enteric bacterial pathogens possess distinctive mechanisms to attenuate host inflammatory responses, which is a prerequisite for promoting intracellular and extracellular bacterial survival [4,5]. For example, during the invasion of the intestinal epithelium by Shigella, the Toll-like receptors (TLRs) and Nod-like receptors (NLRs) are the PRRs that recognize PAMPs and DAMPs [6][7][8][9]. These receptors stimulate host inflammatory signaling pathways, including nuclear factor kB (NF-kB) and mitogen activated protein kinases (MAPK), which culminate in the expression of proinflammatory chemokines and cytokines [6][7][8][9]. Shigella counteract innate immune responses by delivering effector molecules using its type III secretion system (T3SS) [8][9][10]. The targeting molecules and mechanisms of inhibition of the NF-kB pathway are specific to each of the effectors, and Shigella are known to deliver OspG [11], OspI [12], OspZ [13], and IpaH9.8 [14][15][16][17], to efficiently attenuate NF-kB activation thus allowing replication within the intestinal epithelium [8,9].
The Shigella flexneri strain, YSH6000, has three ipaH genes (ipaH9.8, ipaH7.8, and ipaH4.5) on its large virulence plasmid and seven ipaH genes on its chromosome (ipaH0722, ipaH0887, ipaH1383, ipaH1880, ipaH2022, ipaH2202, and ipaH2610) [26][27][28]. We previously showed that all of these IpaH effector proteins are secreted via the T3SS [28,29], suggesting they have the potential to act as E3 ubiquitin ligase effectors during infection, although their exact roles and host targets during infection, with the exception of IpaH9.8, remain largely unclear [17]. In the mouse lung infection model, we showed that mice infected with the DipaH-chromosome mutant, which lacks all seven chromosomal ipaH genes, had a more severe inflammatory response and less colonizing bacteria compared to the WT strain [28], suggesting that modulation of the host inflammatory response by chromosomal IpaH proteins promotes bacterial infection. Therefore, to gain further insight into the strategies employed by Shigella to counteract host innate defence systems during infection, we sought to characterize the chromosomally encoded IpaH effectors. We found that IpaH0722 targets TRAF2 and plays an important role in dampening the PKC-NF-kB pathway, in response to membrane damage generated during Shigella invasion of epithelial cells.

IpaH0722 dampens NF-kB activation during Shigella infection
The infection of mice using the DipaH-chromosome mutant, which lacks all of the chromosomal ipaH genes, leads to an increased production of the NF-kB responsive gene MIP-2 [28]. In this current study, we used HeLa cells to determine the effect of the DipaH-chromosome mutant on the NF-kB activation by measuring the kinetics of IkBa degradation. HeLa cells were infected with YSH6000 (S. flexneri WT) or DipaH-chromosome mutant, and then whole cell lysates were harvested at 20, 40, 60, and 80 min post-infection for the detection of IkBa. The degradation rate of IkBa in cells infected with the DipaHchromosome mutant was higher than that of WT, indicating that one or more of the chromosomal IpaH proteins contributed to the dampening of IkBa degradation (Fig. 1A, left). To identify the IpaH proteins that were involved in suppressing NF-kB activation, we generated deletions mutants for each of the seven chromosomal ipaH genes. IpaH0722 was critical for the inhibition of NF-kB activation. The degradation rate of IkBa was higher in DipaH0722 infection compared to WT infection (Fig. 1A, right). IpaH0722, which corresponds to ORF SF0722 of Shigella flexneri 2a Sf301 strain, has a Cys residue in its C-terminal region that is required for E3 ubiquitin ligase activity [16,28,29].
In Cos-7 and HeLa cells, ectopically-expressed IpaH0722 localized to the cell membrane ( Fig. 1B and S1A). A mouse lung infection model was used to evaluate the role of IpaH0722 in the pathogenesis of Shigella. The survival rates of mice infected with the DipaH0722 mutant Shigella were increased compared to mice infected with WT Shigella (Fig. 1C). Next, to investigate whether the Cys379 residue in the IpaH0722 C-terminal region and its E3 ubiquitin ligase activity contributed to the pathogenesis of Shigella, we substituted Cys379 residue (IpaH0722CA). When 293T cells expressing IpaH0722 or IpaH0722CA were infected with Shigella, the activation of NF-kB was decreased in the presence of IpaH0722 but not IpaH0722CA (Fig. 1D). In addition, to confirm the importance of the E3 ubiquitin ligase activity of IpaH0722 in Shigella infection, HeLa cells were infected with Shigella DipaH0722 harboring ipaH0722 (DipaH0722/ipaH0722) or ipaH0722CA (Di-paH0722/ipaH0722CA). In cells infected with the WT strain or DipaH0722/ipaH0722 strain, the degradation rate of IkBa was reduced at 60 and 80 min after infection compared to the DipaH0722 or DipaH0722/ipaH0722CA strains (Fig. 1E). It is important to note that IpaH0722 had no effect on Elk-1 and JNK activation ( Fig. 1D and E). We speculated that IpaH0722 plays a role in the inhibition of Shigella-induced NF-kB activation in an E3 ubiquitin ligase-dependent manner.
IpaH0722 selectively inhibits phorbol myristate acetate (PMA)-induced NF-kB activation NF-kB activity can be stimulated by multiple signaling pathways that are triggered by various receptors in response to exogenous and endogenous stimuli, including TNFa-TNFR, TLRs, and NLRs [30]. To characterize IpaH0722-dependent dampening of NF-kB activation, we measured the levels of NF-kB activation in IpaH0722-or IpaH0722CA-expressing 293T cells that were stimulated with PMA, TNF-a, lipopolysacharide (LPS), or IL-1busing NF-kB-luciferase reporter assays. IpaH0722 markedly impaired PMA-dependent, but not TNF-a-, LPS-, and IL-1bdependent NF-kB activation in an E3 ubiquitin ligase activitydependent manner ( Fig. 2A). To confirm the specificity of IpaH0722-mediated induction of NF-kB activation, we investigated the expression of IL-8 in 293T cells stimulated with PMA using an IL-8 reporter assay. Consistent with the NF-kB findings, IpaH0722 preferentially inhibited the PMA-dependent IL-8 expression (Fig. 2B). Moreover, the preferential targeting of PMA-but not TNF-a-induced NF-kB activation was also demonstrated by ELISA assays (Fig. 2C). Although PMA was also known to induce MAPK activation, IpaH0722 failed to inhibit PMA-dependent AP-1 and Elk-1 activation (Fig. 2D). Taken together these results clearly indicated that IpaH0722 preferentially inhibited PMA-induced NF-kB activation by targeting factors that modulated the NF-kB signaling pathway.

Ectopically-expressed IpaH0722 localizes to the cell membrane
It has recently reported that Salmonella SspH2, an IpaH cognate protein, localizes to the host cell membrane through its modification by cellular palmitoyl transferases. SspH2 has a putative S-palmitoylation motif in its N-terminal and undergoes modification by host cell palmitoyltransferases, resulting in its localization to the host cell membrane [31]. Since we found a

Author Summary
In response to bacterial infection, host cells induce a plethora of innate immune responses to combat the infection. However, many bacterial pathogens have developed sophisticated mechanisms to evade the host's immune system. Because NF-kB is crucial for innate immune responses against bacterial infection, bacterial pathogens deploy multiple countermeasures to inhibit NF-kB activation. The invasion and replication of Shigella within host cells results in cellular damage and the production of bacterial components that trigger NF-kB activation. Here, we show that the Shigella type III secretion system (T3SS) effector IpaH0722, a member of the IpaH E3 ubiquitin ligase family, inhibits NF-kB activation during Shigella infection. IpaH0722 preferentially targets the PKC-NF-kB pathway, which is activated in response to danger signals caused by disruption of the phagosomal membrane during the dissemination of Shigella into the cytoplasm. IpaH0722 inhibits NF-kB activation by targeting TRAF2, which lies downstream of PKC, for ubiquitination and proteasome-dependent degradation. putative S-palmitoylated motif in IpaH0722 N-terminal portion, we investigated whether this motif played a role in IpaH0722 localization to the cell membrane. In Cos-7 and HeLa cells, ectopically-expressed IpaH0722 localized to host cell membrane ( Fig. 1B and S1A). IpaH0722 has two Cys residues in its palmitoylation motif (Cys14 and Cys18), we generated constructs in which the Cys residues were substituted with Ser singly and in tandem (IpaH0722-C14S, -C18S, and -C14S/C18S). These constructs were ectopically expressed in HeLa cells to evaluate the subcellular localization of IpaH0722. IpaH0722 WT and IpaH0722CA localized to the cell membrane, whereas IpaH0722-C14S, -C18S, and -C14S/C18S accumulated within the cytosol. Therefore, palmitoylation at Cys14 and Cys18 residues of IpaH0722 are critical for its localization to the plasma membrane ( Fig. S1A).
We next investigated whether membrane localization of IpaH0722 was required for its inhibition of NF-kB. The various IpaH0722 constructs were ectopically expressed in 293T cells, which were stimulated with PMA, NF-kB activity was assessed using NF-kB-luciferase reporter assays. IpaH0722, but not IpaH0722CA, inhibited NF-kB activation (Fig. S1B). Moreover, since the disruption of the putative palmitoylation sites in IpaH0722 mutants did not alter their ability to reduce PMAdependent NF-kB activation, we presumed that the localization of IpaH0722 to the plasma membrane was not required for NF-kB inhibition in our experimental setting (Fig. S1B).

Membrane rupture of epithelial cells by Shigella invasion triggers PKC-NF-kB activation
PMA mimics the role of diacylglycerol (DAG) in the activation of the protein kinase C (PKC)-NF-kB pathway [32]. Therefore, we hypothesized that IpaH0722 selectively targeted the DAG-PKC-NF-kB pathway, which was likely due to the membrane localization of IpaH0722 once it was secreted by invading Shigella into epithelial cells. We sought to determine which of the PKC isoforms were involved in the activation of NF-kB. The PKC family of proteins is composed of 12 isoforms that act as lipidactivated Ser/Thr kinases [33]. The PKC family can be divided into four functional protein classes: conventional PKC (PKCa, b, and c), novel PKC (PKCd, e, g, and h), atypical PKC (PKCf, l/i) and PKC related kinase (PKCm). Conventional and novel PKCs have a DAG binding domain. Conventional PKCs require DAG and Ca 2+ for their activation, whereas novel PKCs require only DAG for their activation. Atypical PKCs do not depend on DAG and Ca 2+ for their activation. Since the activity of PKC is regulated by phosphorylation and its recruitment to the cell membrane, we investigated the levels of phosphorylated PKC during Shigella infection of HeLa cells. HeLa cells were infected with Shigella WT and cell lysates were harvested at 10, 20, 40, and 60 min after infection for the analysis of phosphorylated PKC by immunoblotting. As shown in Fig. 3A, Shigella infection augmented the phosphorylation of conventional or novel PKCs, such as PKCd, at 10 and 20 min, and PKCm at 20, 40, and 60 min postinfection.
To investigate whether PKC activation triggers NF-kB activation in Shigella infection, we exploited dominant negative forms (DN) of PKC and siRNA that targeted PKC. To investigate the role of PKC in the activation of NF-kB during Shigella infection, DN-PKCa and DN-PKCd, e, h, were ectopically expressed in 293T cells. The DN-PKCd, e, h, but not DN-PKCa, significantly decreased Shigella-induced NF-kB activation (Fig. 3B) suggesting that the PKCd-NF-kB pathway plays a critical role during Shigella infection. Similarly, siRNA-mediated knockdown of PKCd in 293T cells infected with Shigella decreased NF-kB activity to less than half of the control levels ( Fig. 3C).
The invasion of epithelial by Shigella produces membrane ruffles through the remodeling of the F-actin cytoskeleton [34][35][36]. Immediately following bacterial invasion, the bacteria are rapidly surrounded by a vacuolar membrane that the bacteria disrupt to facilitate dissemination into the cytoplasm by inducing actin polymerization [34][35][36]. To further understand the mechanism of PKC activation by Shigella invasion, we examined PKC phosphorylation during Shigella infection. First we confirmed the importance of Shigella invasiveness. When HeLa cells were infected with Shigella WT or the T3SS deficient mutant S325, Shigella WT infection triggered the phosphorylation of PKCd, however the S325 mutant did not (Fig. 3D). We next sought to determine whether vacuolar membrane disruption potentiated DAG-PKC-NF-kB signaling. We previously created an DipaB/inv mutant by introducing Yersinia invasin gene into the DipaB mutant [37]. The resulting DipaB/inv mutant was internalized into the endocytic vacuole and it was unable to disrupt the vacuolar membrane for dissemination into the cytoplasm (Fig. 3D). When HeLa cells were infected with Shigella WT or the DipaB/inv mutant, Shigella WT infection triggered the phosphorylation of PKCd, however the DipaB/inv mutant did not (Fig. 3D). In contrast, both WT and DvirG mutant, which is unable to support intra-and inter-cellular movement, induced PKCd phosphorylation (Fig. 3D). To further confirm the importance of vacuolar membrane disruption, we investigated PKC phosphorylation during Salmonella infection using a strain that triggers membrane ruffling but remain sequestered in the phagosome. When HeLa cells were infected with Shigella WT or the Salmonella WT, Shigella WT but not Salmonella WT infection triggered the phosphorylation of PKCd (Fig. 3D). These data supported the notion that membrane rupture by Shigella in infected epithelial cells triggers the activation of the DAG-PKC-NF-kB pathway.

IpaH0722 inhibits the PKC-NF-kB pathway without targeting the CBM complex
Because Shigella invasion of epithelial cells triggers the DAG-PKC-NF-kB signaling, we sought to determine the host factors in the PKC-NF-kB signaling pathway that were targeted by IpaH0722 E3 ubiquitin ligase. First, we determined which of the steps during the activation of PKC-NF-kB was targeted by plasma membrane. Cos-7 cells were transfected with a Myc-IpaH0722 expression vector and immunostained with anti-Myc, actin, and TO-PRO3. (C) The murine pulmonary model of Shigella infection. Mice were intranasally inoculated with Shigella WT or DipaH0722 (n = 13) at 5610 7 cfu and survival was recorded for 10 days post-infection. *P,0.05. (D) NF-kB-luciferase and Elk-1-luciferase reporter assays were performed after Shigella infection of 293T cells that were transiently transfected with empty vector, IpaH0722, or IpaH0722CA expressing plasmids. Results are presented as 'fold' relative to the activity of uninfected or unstimulated cells. *P,0.01, n.s., not significant. (E) HeLa cells were infected with Shigella WT, DipaH0722, or DipaH0722 harboring ipaH0722 or ipaH0722CA. The cell lysates prepared at the indicated time points were subjected to immunoblotting. Anti-actin antibody served as the loading control. doi:10.1371/journal.ppat.1003409.g001 IpaH0722. The NF-kB activity induced by PKC signaling factors was measured in 293T cells with ectopic expression of IpaH0722 or IpaH0722CA. The transiently transfected cells were stimulated with a subset of constitutively active PKC isoforms (PKCa, -d, -e, and h) and NF-kB activity was determined. IpaH0722, but not IpaH0722CA, inhibited all PKC isoforms (PKCa, -d, -e, and -h) and NF-kB activation, indicating that IpaH0722 targeted PKC itself or factors that lie directly downstream of PKC (Fig. 4A). In HeLa cells, similar levels of phosphorylated PKCd were detected in response to Shigella WT and DipaH0722 infection (Fig. 4B). Moreover, no differences were detected in PKC phosphorylation in PMA plus ionomycin stimulated 293T cells expressing IpaH0722 or IpaH0722CA (Fig. 4C). In addition, IpaH0722 did not bind to PKC, and had no effect on protein stability of PKC, suggesting that IpaH0722 targets downstream of PKC rather than PKC itself (Fig. S2). We therefore focused on the CARMA-Bcl10-MALT1 (CBM) complex because the phosphorylation of CARMA1, 2, and 3 by PKC induces a conformational change that recruits Bcl10-MALT1 to CARMA (CBM signalosome), which is essential for the activation of downstream signaling [38,39]. Therefore, we tested the effect of IpaH0722 on CBM complex formation, and found that IpaH0722 failed to inhibit CARMA1-, CARMA2-, CARMA3-, or Bcl10-induced NF-kB activity (Fig. 4D). Indeed, IpaH0722 did not bind to CARMA1-, CARMA2-, CARMA3-, or Bcl10, and had no effect on their protein stability (Fig. S3). Furthermore, since IpaH0722 did not affect PKC-mediated CARMA phosphorylation (Fig. 4E), we believe that IpaH0722 inhibited the PKC-NF-kB pathway but not via the CBM complex.

IpaH0722 preferentially inhibits TRAF2-mediated NF-kB activation
We subsequently investigated other NF-kB signaling factors, namely TRAF2, TRAF5, TRAF6, NIK, IKKa, IKKb, or p65 as potential IpaH0722 targets [40]. To this end, 293T cells expressing NF-kB-luciferase reporter gene together with TRAF2, TRAF5, TRAF6, NIK, IKKa, IKKb, or p65 were transfected with a vector encoding IpaH0722 or IpaH0722CA and luciferase activity was measured (Fig. 5A). IpaH0722, but not IpaH0722CA, inhibited NF-kB activity when the cells that expressed TRAF2 suggesting that TRAF2 is a target for IpaH0722 E3 ubiquitin ligase (Fig. 5A). In previous reports, PKC-mediated phosphorylation of TRAF2 was a prerequisite for NF-kB activation in epithelial cells [41,42]. PKCd and PKCe phosphorylated TRAF2 at residue Thr117 resulting in IKK recruitment and ultimately NF-kB activation [41,42]. We therefore used siRNA-mediated TRAF2 knockdown cells to measure the effect of TRAF2 on activation of the PKC-NF-kB pathway. The results showed that TRAF2 knockdown decreased NF-kB activation in response to PMA stimulation and Shigella infection (Fig. S4A). To further confirm the effect of TRAF2 knockdown on NF-kB activation we utilized Traf2-knockout mouse embryonic fibroblasts (MEF) [43]. We transduced Traf2 into the Traf2-knockout MEFs by retrovirus infection to generate stably expressing Traf2 cell lines. The degradation of IkBa in Traf2 stably expressing MEF cells and Traf2-knockout MEF cells during Shigella infection was determined. As shown in Fig. S4B, the degradation rate of IkBa in Traf2 stably expressing MEFs was higher than that of Traf2knockout MEFs at 20 and 40 min after infection suggesting that TRAF2 plays a role in Shigella-induced NF-kB activation.
In previous reports, structural analysis of IpaH proteins using IpaH1.4 or IpaH3 revealed that the LRR region of the IpaH family is required for substrate recognition, while the CTR region is essential for its E3 ubiquitin ligase activity [44,45]. IpaH proteins share characteristic domains: an N-terminal 60-70 amino acid stretch, a LRR region, and an intervening region flanked by the LRR and the conserved CTR (Fig. S5A). We constructed a series of GST-tagged IpaH0722 truncations and performed GSTpull down assays using HeLa whole cell lysates. IpaH0722 truncations that contained the LRR regions bound to TRAF2; however, IpaH0722 truncations that contained the N-terminal, intervening region, or CTR failed to bind to TRAF2 (Fig. S5B).

IpaH0722 targets TRAF2 for ubiquitination and proteasomal degradation
Since IpaH0722 inhibited NF-kB activation in an E3 ubiquitin ligase-dependent manner, we tested whether IpaH0722 ubiquitinated TRAF2 using an in vitro ubiquitination assay. TRAF2, IpaH0722, or IpaH0722CA (purified from E. coli) were combined with E1, ATP, UbcH5b (an E2 ubiqitin conjugating enzyme) in assay medium and subjected to immunoblotting. TRAF2 was ubiquitinated by IpaH0722 but not IpaH0722CA (Fig. 6A). To confirm the fate of ubiquitinated TRAF2, we measured the stability of TRAF2 in 293T cells ectopically expressing IpaH0722 or IpaH0722CA and treated with cycloheximide (CHX; a protein synthesis inhibitor) for 2, 4, and 6 h. TRAF2 degradation was faster in IpaH0722 expressing cells compared to IpaH0722CA expressing cells, which confirmed that IpaH0722 targeted TRAF2 for ubiquitin-mediated protein degradation (Fig. 6B). We also measured TRAF2 degradation in 293T cells that were co- Figure 2. IpaH0722 selectively inhibits PMA-induced activation of the NF-kB pathway. (A) Luciferase reporter assays of 293T cells transiently transfected with an NF-kB reporter plasmid and empty vector, IpaH0722, or IpaH0722CA expressing plasmids. After 24 h, cells were treated with TNF-a, PMA, LPS, or IL-1b for 3 h and luciferase activity was measured. Results are presented as fold change relative to the activity of uninfected or unstimulated cells. *P,0.01 (B) IL-8 expression in 293T cells. Cells expressing empty vector, IpaH0722, or IpaH0722CA expressing plasmids were treated with TNF-a, PMA, or IL-1b for 3 h. Measurement of IL-8 reporter activity after PMA, TNF-a, and IL-1b stimulation. Results are presented as fold change relative to the activity of uninfected or unstimulated cells. *P,0.01. (C) The levels of IL-8 production were measured by ELISA. Cells expressing empty vector, IpaH0722, or IpaH0722CA expressing plasmids were treated with PMA and TNF-a, then supernatants were harvested for ELISA analysis. *P,0.0. (D)Luciferase reporter assays of 293T cells transiently transfected with NF-kB, AP-1, or Elk-1 reporter plasmids plus empty vector, IpaH0722, or IpaH0722CA expressing plasmids. After 24 h, cells were treated with PMA for 3 h and luciferase activity was measured. Results are presented as fold change relative to the activity of uninfected or unstimulated cells.*P,0.01, n.s., not significant. doi:10.1371/journal.ppat.1003409.g002 transfected with TRAF2 plus IpaH0722 or IpaH0722CA, and treated with MG132 (a proteasome inhibitor) or E64D plus pepstain A (lysosome inhibitors). As shown in Fig. 6C, MG132 treatment prevented TRAF2 degradation in the presence of IpaH0722, whereas E64D plus pepstatin A treatment did not prevent TRAF2 degradation. These data confirm that IpaH0722 targets TRAF2 for ubiquitination and proteasomal degradation.
Previous studies indicated that TRAF5 compensates for TRAF2 function [43], so we exploited Traf2/Traf5 double knockout MEFs to clarify the requirement of TRAF2 on NF-kB activation. We introduced Traf2 into the Traf2/Traf5 double knockout MEFs by retrovirus transduction. Under puromycin selection, we obtained stable TRAF2-expressing Traf2/Traf5 double knockout MEFs. Traf2/Traf5 double knockout MEFs and stable Traf2-expressing Traf2/Traf5 double knockout MEFs were infected with Shigella WT or DipaH0722, and IkBa degradation was measured over time.
The degradation rate of IkBa in DipaH0722-infected stable TRAF2-expressing Traf2/Traf5 double knockout MEFs was higher compared to WT Shigella infection (Fig. 6D). In contrast, the degradation rates of IkBa in WT and DipaH0722-infected double knockout MEFs were almost equal (Fig. 6D). To evaluate the importance of the in vivo role of E3 ligase activity of IpaH0722 for Shigella virulence, we infected mice with a sublethal dose of Shigella WT, DipaH0722, DipaH0722/0722, or DipaH0722/ 0722CA. Mice infected with DipaH0722 or DipaH0722/0722CA mutants showed significantly reduced bacterial colonization when compared to that of WT or DipaH0722/0722 infection (Fig. S6A). Furthermore, histopathology showed that inflammatory responses, such as suppurative masses, neutrophil infiltration, and macrophage infiltration, were significantly higher in mice infected with DipaH0722 or DipaH0722/0722CA mutants when compared to that of WT or DipaH0722/0722 infections (Fig. S6B). These results supported our notion that the interaction between IpaH0722 and TRAF2 is an important mechanism for dampening NF-kB activation during Shigella infection of epithelial cells.

Discussion
In this study we show that IpaH0722, a chromosomallyencoded IpaH E3 ligase, specifically dampens the PKC-dependent activation of NF-kB in response to Shigella invasion of epithelial cells. We find the IpaH0722 exerts its effect by targeting TRAF2 for proteasomal-dependent degradation (Fig. 6E). The inhibitory activity possessed by IpaH0722 appears to be an important mechanism for the downregulation of the acute inflammatory response by blocking the activation of PKC-NF-kB pathway. This is a critical step since cell membrane rupture is a part of the process of Shigella invasion into epithelial cells, and the recognition of this as a DAMP triggers the recruitment and activation PKC, and ultimately the activation of the innate immune response via PKC-NF-kB pathway. This notion was supported by the experiments using the DipaB/inv mutant, which was internalized by the vacuolar membrane but could not rupture the vacuole [36], we showed that Shigella WT infection, but not DipaB/inv mutant infection, caused in the phosphorylation of PKCd (Fig. 3E). In fact, Dupont et al. recently reported that the vacuolar membrane remnants generated by Shigella invasion into epithelial cells undergo ubiquitination and provided a cue to stimulate autophagy and inflammatory responses [46]. Moreover, Shahnazari et al. showed that the DAG generated in Salmonella containing vacuole (SCV) membranes serves as a specific signal to induce autophagy through recruitment of PKC into the SCV [47]. Tattoli et al. showed that vacuolar membrane damage triggers amino acid starvation and the downregulation of mTOR activity, thereby inducing autophagy against intracellular Shigella and Salmonella [48]. Phagosomal membrane disruption by Listeria monocytogenes LLO in macrophages, which is known as a cholesterol-dependent cytolysin, triggers the recruitment of PKC to the membrane and its activation [49]. After disruption of the Listeria phagosome the membrane damage was recognized by PKC. These data suggest that PKC functions as a part of the repair mechanism of host cells, and PKC recruitment by the membrane rupture is recognized as a DAMP by the innate immune system [49]. These studies also imply that the membrane rupture, which is created by bacterial infection, acts as a double-edged sword for the pathogens; on one hand it benefits the bacteria by promoting further dissemination and multiplication, and on the other it hinders the bacteria by acting as a DAMP that stimulates the host innate immune response. In line with this, we identified IpaH0722 as a Shigella effector that plays an important pathogenic role because it counteracts DAMPs-mediated innate immune responses.
Recently we reported that another Shigella effector, OspI targets UBC13 and dampens the DAG-CBM-TRAF6-NF-kB pathway, which is triggered by membrane ruffles around the Shigella entry sites into epithelial cells [12]. OspI acts as a glutamine deamidase for UBC13 to convert Q100 to E100, thus inactivating UBC13 E2 ubiquitin conjugating enzymatic activity that is required for TRAF6 activation [12]. Currently, although the reason underlying why Shigella deliver IpH0722 in addition to OspI during the invasion of epithelial cells remains partly unclear, we believe that it is due to the fact that the two different signaling pathways are required to neutralize the acute inflammatory response and are triggered by distinct bacteria-induced cellular events: DAG-CBM-TRAF6-NF-kB (membrane ruffles) and DAG-PKC-TRAF2-NF-kB (membrane rupture) [12; this study].
TRAF2 and TRAF5 have been extensively studied in TNFR-NF-kB signaling in immune cells, and it has also been reported that TRAF2 is essential in Nod1-, PKC-, and non-canonical NF-kB signaling in non-myeloid cells [42,50,51]. In previous studies, PKC-mediated phosphorylation of TRAF2 at Ser11, Ser55, and Thr117 was required for prolonged NF-kB activation, but not JNK activation [42,[52][53][54]. Consistent with these observations, our data showed that treatment of siRNA-mediated knockdown of Traf2 decreased PMA-induced NF-kB activation, suggesting that TRAF2 is involved in PKC-NF-kB signaling (Fig. S4). According to Hasegawa et al., Traf2/Traf5 double knockout MEFs but not TRAF6 knockout MEFs had decreased NF-kB activation upon Nod1-RIP2 stimulation, suggesting that TRAF2 was also important in the Nod1-NF-kB pathway [51]. Since Nod1 recognizes the peptidoglycan that is released from Shigella at the entry site and in the cytoplasm, and plays an important role in Shigella-induced NF-kB activation, IpaH0722 likely inhibits both the PKC-NF-kB pathway and the Nod1-RIP2-TRAF2-NF-kB pathway during Shigella infection [55][56][57]. In fact, we see that IpaH0722 also inhibited Nod1-stimulated NF-kB activation in our reporter assay (data not shown). We have provided evidence that IpaH0722 interacted with TRAF2 via its substrate recognition site, LRR, and modified its ubiquitination ( Fig. 6 and Fig. S5). The ubiquitination of TRAF2 by IpaH0722 led to its proteasomal-dependent degradation and inhibited NF-kB activation (Fig. 6).
Though not yet fully understood, IpaH0722 does not inhibit NF-kB in response to TNF-a; unidentified conformational changes or signaling pathways involving TRAF2 likely occur during Shigella infection or PMA stimulation.
Herein, we provide the first evidence that Shigella invasion induced membrane rupture of epithelial cells, which stimulates PKC-NF-kB signaling. Moreover, Shigella used the E3 ubiqitin ligase effector, IpaH0722, to target TRAF2 for proteasomaldependent degradation (Fig. 6E). This is the first report IpaH0722 as a non-plasmid-encoded T3SS effector protein in Shigella.

Cell culture
HeLa cells were cultured in Eagle's minimal essential medium (Sigma) supplemented with 10% fetal calf serum. Cos-7 and 293T cells were cultured in Dulbecco's modified Eagle medium (Sigma) supplemented with 10% fetal calf serum. The Traf2-deficient or Traf2/Traf5-deficient MEFs were maintained as described previously [43]. To construct Traf2 2/2 or Traf2/Traf5 2/2 cells stably expressing Traf2, cDNAs encoding these genes were subcloned into pMX-puro retroviral expression vectors. Retroviral supernatants were produced in Plat-E cells. Target cells were transduced with supernatants in the presence of DO-TAP (Roche) and then cloned under puromycin selection.

Bacterial infection
HeLa cells were infected with various strains of Shigella at a multiplicity of infection (moi) of 100. In the case of the Shigella strains expressing afimbrial adhesin (Afa), cells were infected at moi of 10. To adjust the intracellular bacterial number, DipaB/inv were set to 3 times moi of WT. Infection was initiated by centrifuging the plate at 7006 g for 10 min. After incubation for 20 min at 37uC, the plates were washed three times with PBS, transferred into fresh medium containing gentamicin (100 mg/ml) and kanamycin (60 mg/ml) to kill extracellular bacteria. Salmonella infection was conducted as described previously [58]. After incubation for indicated time, the cells were washed with PBS and harvested into 26 Laemmli's sample buffer for immunoblotting. The density of each band was quantified by measuring the mean intensity using NIH image software version 1.63 and the expression levels were normalized to the levels of b-actin. All immunoblotting experiments were repeated multiple times and representative data of similar results are presented.

Luciferase reporter assays
293T cells were seeded into 24-well plates. After 24 h, the cells were transfected with reporter plasmids (pNF-kB-luc, pIL-8-luc, pElK-1-luc, or pAP-1-luc), Renilla luciferase constructs (phRL-TK, Promega) using FuGENE6 transfection reagent (Roche). The IL-8 reporter plasmid was constructed by amplifying the 181 bp human IL-8 promoter sequence from the genomic DNA of HeLa cells, which was then inserted into the pMet-luc reporter plasmid (Clontech). Equal amounts of empty vector were used to control for the transfection process. After 24 h, cells were infected with Shigella WT (moi = 30) or treated with PMA (25 nM), TNF-a (10 ng/ml) or LPS (100 ng/ml) for 3 h. To investigate LPS stimulation, TLR4 and MD2 vectors were also transfected into the cells. Cell extracts were prepared and reporter activity was determined using the luciferase assay system (Promega). Results are presented as fold-change relative to the activity of uninfected or unstimulated cells. Data are representative of three independent experiments.

Expression and purification of recombinant proteins
For GST fusion proteins, E. coli BL21 (DE3) strain harboring pGEX4T-1 derivatives were cultivated in L-broth supplemented with ampicillin (50 mg/ml) 3 h at 30uC. Expression was induced by the addition of 1 mM IPTG and incubation for 3 h at 30uC. Bacteria were disrupted by sonication and lysozyme treatment. Purification of the GST fusion proteins with glutathione-Sepharose 4B (GE Healthcare) was performed according to the manufacturer's protocol.

Pull-down assay
GST-IpaH0722 bound to glutathione sepharose 4B beads was mixed with HeLa cell lysates for 2 h at 4uC. After centrifugation, the beads were washed five times with 0.5% Triton X-100-PBS and subjected to immunoblotting. The immunoblotting experiments were repeated multiple times and representative data of similar results are presented.

Immunoprecipitation
293T cells were transiently transfected using FuGENE 6 (Roche). Cells were washed with PBS and lysed for 30 min at 4uC in lysis buffer containing 150 mM NaCl, 50 mM HEPES pH 7.5, 1 mM EDTA, 0.5% NP-40, and Complete protease inhibitor cocktail (Roche). Lysates were cleared by centrifugation and proteins were immunoprecipitated 2 h with anti-Myc (9B11) or anti-M2 FLAG and Protein G beads (Sigma) at 4uC. Immunoprecipitates were washed five times with lysis buffer and subjected to immunoblotting. The immunoblotting experiments were repeated multiple times and representative data of similar results are presented.

In vitro ubiquitination assays
In vitro TRAF2 ubiquitination assays were performed in 40 ml reaction mixture containing reaction buffer (25 mM Tris-HCl [pH 7.5], 50 mM NaCl, 5 mM ATP, 10 mM MgCl 2 , and 0.1 mM DTT), 1 mg TRAF2, 0.5 mg E1, 2 mg UbcH5b, and 2 mg ubiquitin purified from E. coli in the presence and absence of GST-IpaH0722 or GST-IpaH0722CA. Reactions were incubated at 37uC for 1 h and stopped by the addition of 56 Laemmli sample buffer. All immunoblotting experiments were repeated multiple times and representative data of similar results are presented.

Cycloheximide chase assay
At 24 h after transfection, 293T cells were treated with cycloheximide (50 mg/ml) (Wako) for the indicated times and cell lysates were harvested for immunoblotting. The density of each band was quantified by measuring the mean intensity using NIH image software version 1.63. The expression levels were normalized to the levels of b-actin. All immunoblotting experiments were repeated multiple times and representative data of similar results are presented.

Murine pneumonia model
The pulmonary infection model in the mice was described previously [28]. In brief, five-week-old female Balb/c mice (CREA Japan) were housed in the animal facility of the Institute of Medical Science, University of Tokyo, in accordance with University guidelines. S. flexneri were suspended in sterile saline, and 20 ml of the bacterial suspension was administered intranasally at 1610 8 cfu (survival assay) or 1610 7 cfu (colonized bacterial number assay). For histological analysis, mice were sacrificed at 24 h after infection, and their lungs were harvested, fixed in 4% paraformaldehyde in PBS, and frozen in liquid nitrogen for sectioning. The sections were stained with hematoxylin and eosin and examined under a microscope. Viable bacteria in the lung tissue were counted by culturing homogenized tissue for 18 h on LB agar plates. Each data point is the mean of the values for 6 infected mice in each group.

Ethics statement
All animal experiments were carried out in strict accordance with the University of Tokyo's regulations for Animal Care and Use protocol, which was approved by the Animal Experiment Committee of the Institute of Medical Science, the University of Tokyo (approval number; PA11-92). The committee acknowledged and accepted both the legal and ethical responsibility for the animals, as specified in the Fundamental Guidelines for Proper Conduct of Animal Experiment and Related Activities in Academic Research Institutions under the jurisdiction of the Ministry of Education, Culture, Sports, Science and Technology, 2006 (Japan). All surgery was performed under carbon dioxide euthanasia, and all efforts were made to minimize suffering.

Statistical analysis
Values are reported as means 6 standard deviation (SD) or means 6 standard error (SEM) of data obtained for independent experiments. Statistical analysis was performed Student's t-test or a one-way ANOVA. Kaplan-Meier survivor curves were generated. p-values,0.05 were considered significant.