The Type III Secretion Effector NleE Inhibits NF-κB Activation

The complex host-pathogen interplay involves the recognition of the pathogen by the host's innate immune system and countermeasures taken by the pathogen. Detection of invading bacteria by the host leads to rapid activation of the transcription factor NF-κB, followed by inflammation and eradication of the intruders. In response, some pathogens, including enteropathogenic Escherichia coli (EPEC), acquired means of blocking NF-κB activation. We show that inhibition of NF-κB activation by EPEC involves the injection of NleE into the host cell. Importantly, we show that NleE inhibits NF-κB activation by preventing activation of IKKβ and consequently the degradation of the NF-κB inhibitor, IκB. This NleE activity is enhanced by, but is not dependent on, a second injected effector, NleB. In conclusion, this study describes two effectors, NleB and NleE, with no similarity to other known proteins, used by pathogens to manipulate NF-κB signaling pathways.


Introduction
Enteropathogenic Escherichia coli (EPEC) belong to a group of pathogens defined by their ability to form ''attaching and effacing'' (AE) histopathology on intestinal epithelia. These pathogens employ their type III protein secretion system (TTSS) to inject (translocate) toxic proteins (effectors) into the host cell. The injected effectors subvert normal host cell functions to benefit the bacteria (summarized in [1]). To date, 21 effectors or putative effector genes have been described for EPEC. Six of them are encoded in the LEE region that also encodes the TTSS structural genes, whereas the other 15 effector genes are distributed within three prophages and three insertion elements (IE) [2].
Upon infection, bacterial PAMPs (pathogen-associated molecular patterns) including LPS, flagellin, lipoproteins, and CpG DNA stimulate host cell Toll-like receptors (TLRs) in the host cells, leading to a formidable immune response via the activation of the transcription factor NF-kB [3,4]. NF-kB comprises a family of closely related transcription factors that play a key role in the expression of genes involved in inflammation, immune, and stress responses. NF-kB is a collective term used for homoand heterodimeric complexes formed by the Rel/NF-kB proteins. In mammals, five of such proteins are known: RelA (p65), RelB, c-Rel, p50 (NF-kB1), and p52 (NF-kB2). Under nonstimulating conditions, NF-kB is retained in the cytoplasm through its association with inhibitory proteins (IkBs). A variety of signaling pathways activate IkB kinases (IKK) to phosphorylate IkB, leading to its ubiquitination and degradation by the proteasome. This allows translocation of NF-kB to the nucleus, activation of NF-kB-regulated genes, and establishment of an inflammatory response [5,6].
Previous reports have suggested that during infection, EPEC manipulate NF-kB-mediated inflammation. Initially, it was shown that EPEC activate NF-kB by a TTSS-dependent mechanism [7,8], but later, it was reported that the TTSS is not required and that EPEC activate NF-kB via a TTSS-independent mechanism, presumably by activation of TLRs [9,10,11,12]. Moreover, some reports showed that EPEC actually repress NF-kB activation by a TTSS-dependent mechanism [13,14]. Taken together, these reports suggest that EPEC first elicit NF-kB activation by a TTSS-independent mechanism and subsequently, it utilize the TTSS mechanism to mediate TTSS-dependent NF-kB-repression. However, the major gap in the above hypothesis is that the putative effector that presumably represses NF-kB activation has never been identified. In this report we confirm that EPEC block NF-kB activation via a TTSS-dependent mechanism and show that the NleE effector is necessary and sufficient to block NF-kB activation via inhibition of IkB phosphorylation and thus induces its stabilization. In addition, we show that a second effector, NleB, is required for better repression of NF-kB activation, suggesting that the function of NleB is related to that of NleE.

EPEC inhibit IkB degradation and NF-kB activation by a TTSS-dependent mechanism
The ability of EPEC to either inhibit or induce NF-kB activation is controversial. Therefore, we re-examined this point using HeLa cells as host cells and IkB stability as a read-out for NF-kB activation. Importantly, TNFa treatment strongly stimulates IkB degradation in these cells, but under the used experimental conditions they exhibited minimal IkB degradation upon exposure to the PAMPs of the infecting EPEC. This allowed the uncoupling of infection and NF-kB activation. HeLa cells were infected with EPEC culture for 3 h, during which the bacteria injected the TTSS effectors into host cells. The infected cells were then treated with 10 ng/ml TNFa to activate NF-kB and at different time points post TNFa-induction, cellular lysates were subjected to western analysis. The results show that TNFa treatment induced rapid degradation of IkB in uninfected cells or cells infected with EPEC TTSS-deficient mutant (escN::kan). In contrast, IkB in cells infected with wild-type EPEC remained stable (Fig. 1A).
We next tested whether the stabilization of IkB by EPEC was associated with inhibition of NF-kB translocation to the nucleus. To monitor NF-kB activation, we used a reporter cell line (AGS SIB02) stably expressing the NF-kB subunit p65 fused to GFP. Cells were infected with wild-type EPEC or, as a negative control, with EPEC TTSS mutant (escV::kan). After 3 h of infection, cells were washed, induced with TNFa, and at 15, 45, 60, and 75 min post TNFa-induction, the cells were fixed, stained with Hoechst 33342, and analyzed by automated microscopy. Importantly, whereas wild-type EPEC repressed p65-GFP translocation to the nucleus, the TTSS escV mutant was strongly attenuated in this activity (Fig. 1B). To validate the above microscopic analysis we carried out identical experiment, but instead of using microscopy to analyze the cells we fractionated the cells into cytoplasmic and nucleus fractions and determined the amount of p65 in the different fractions by immunoblot using anti-p65 antibody. The results were in agreement with the microscopic analysis: wild type EPEC, but not the escV mutant, blocked translocation of p65 to the nucleus (Fig. 1C). Thus supporting the notion that EPEC inhibit NF-kB activation by a TTSSdependent mechanism [13,14], and suggesting that EPEC deliver into infected cells one or more effectors that inhibits NF-kB activation.
NleH has been proposed as such an effector since it is similar to OspG, a Shigella effector that inhibits NF-kB activation [15]. However, we found that an EPEC strain, in which both nleH alleles were deleted, still inhibited IkB degradation, similarly to wild-type EPEC (Fig. 1D), suggesting that NleH is not required for blocking IkB degradation under the experimental conditions used by us.

Author Summary
The innate immune system senses intruding pathogens and in response, mounts an inflammatory reaction. Essential for this response is the activation of the transcription factor NF-kB, which mediates reprogramming of gene expression in the host. The bacteria Escherichia coli is usually a nonpathogenic resident of our intestinal track. Some E. coli strains, however, cause disease or food poisoning; one of these pathogenic strains is enteropathogenic E. coli (EPEC). This pathogen employs a syringe-like organelle, termed type three secretion system (TTSS), to inject into the intestinal host cell a battery of toxic proteins termed effectors. We found that two of the effectors that EPEC injects into the host cell upon infection block the activation of NF-kB and thus interfere with the host immune response. These findings elucidate the intricate cross-talk between the host immune system and the pathogen. HeLa cells were infected with wild-type EPEC (WT), a TTSS mutant (escN::kan) or remained uninfected (N/I). After 3h, cells were washed and treated with 10 ng/ml TNFa. At the indicated time post TNFa treatment, cells were extracted and subjected to western blot analysis with anti-IkB antibodies and anti-tubulin (loading control). (B) AGS SIB02 cells expressing p65-GFP were infected with wild-type EPEC or with a EPEC TTSS mutant (escV::kan) or remained uninfected. After 3 h, the cells were washed and treated with TNFa. At the indicated time post TNFa treatment, cells were fixed, nuclei stained, and the levels of p65-GFP translocation in ,200 cells were analyzed by automated microscopy and image analysis software. Standard errors are indicated by bars. (C) HeLa cells were infected with different strains and treated with TNFa as described for (B). At 30 min post TNFa induction, cells were harvested and fractionated into nuclear and cytoplasmic fractions. The presence of the p65 in the different fractions was analyzed by Western blot using anti-p65 antibody. (D) IkB degradation assay was conducted (as described for A) using uninfected cells, cells infected with EPEC wild-type strain and with an EPEC mutant deleted of its two nleH alleles (nleH mutant). doi:10.1371/journal.ppat.1000743.g001 NleE and NleB are required for IkB stabilization To identify putative effector(s) that block IkB degradation, we bioinformaticly compared the genome of EPEC to that of nonpathogenic E. coli K12 and identified large EPEC-specific regions that contain, or possibly contain, effector genes. Based on this comparison, we constructed a set of 15 EPEC strains, each deleted of one EPEC-specific large chromosomal region (Table 1). Altogether, 770 EPEC-specific ORFs were deleted. We then tested the capacity of each of the deleted strains to inhibit IkB degradation upon TNFa treatment. One of the strains, deleted of the IE6 region [2], could not inhibit IkB degradation (data not shown). Further systematic deletion analysis defined two effector-encoding genes, nleB and nleE, required for stabilizing IkB ( Fig. 2A, 2B and data not shown). Deletion of nleE strongly reduced the bacteria's capacity to stabilize IkB, but a complete deficiency in IkB stabilization was observed only in the strain deleted of both nleB and nleE (Fig. 2B).
To corroborate the notion that NleE is required for IkB stabilization, we complemented a strain deleted of the nleBE region with plasmids containing nleB, nleE, or nleBE. We found that expression of NleE, but not of NleB, partially restored EPEC's capacity to stabilize IkB (Fig. 2C). Importantly, full IkB protection was achieved in strains expressing both NleB and NleE (Fig. 2C). A mutant expressing only NleB showed only low level of IkB protection (Fig. 2C). Taken together, these results suggest that NleB and NleE, located at the IE6 region, are necessary for stabilizing IkB and that this activity is contributed mainly by NleE (Fig. 2B). We therefore focused our attention on NleE.
NleE IE2 is not required for IkB stabilization EPEC encode two very similar nleE alleles. One allele, identified in our screen, is located in the IE6 region and the other is in the IE2 region [2]. We initially found that deletion of the IE6 region, but not of the IE2 region, caused deficiency in inhibition of IkB degradation (data not shown). However, the two proteins, NleE IE2 and NleE IE6 , are identical, apart from an internal deletion of 56 residues in NleE IE2 (Fig. S1), and this similarity between the two proteins urged us to determine the activity of each of the two proteins. We first tested their ability to complement IkB destabilization in a strain deleted of nleE IE6 . To this end, we expressed each of them on a plasmid carrying an identical promoter and ribosomal binding site. Results showed that only NleE IE6 , but not NleE IE2 , was able to attenuate IkB degradation (Fig. 3A). These results indicate that NleE IE2 is either not active in the host cell or is not translocated into the host cell. To differentiate between these two possibilities, we used the above mentioned plasmid where both proteins were fused to the blactamase translocation reporter protein, BlaM. The plasmids were introduced into EPEC and the ability to translocate them into infected cells was tested. We found that both NleE IE2 -BlaM and NleE IE6 -BlaM were expressed at similar levels in the bacteria (Fig. 3B). Importantly, however, only NleE IE6 was translocated into the host cell (Fig. 3C), suggesting that NleE IE2 is a cryptic effector. Cumulatively, these results define NleE IE6 , but not NleE IE2 , as the effector needed for inhibition of IkB degradation. Therefore, in this report the term ''NleE'' specifically refers to ''NleE IE6 ''.

NleE is required for complete inhibition of NF-kB translocation to the nucleus
To corroborate the notion that NleE is required for inhibition of NF-kB activation, we examined whether the nleE mutant is deficient in blocking the translocation of the p65 NF-kB subunit to the nucleus upon TNFa treatment. Because of the high similarity between the IE2 region and the IE6 region, we made the specific mutants (DnleE IE6 , DnleEB IE6 ) in a strain deleted of its IE2 region (DIE2). Therefore, these specific strains were compared to their parental strain (indicated as WT DIE2 in the figures). It should be emphasized that the DIE2 mutant exhibited the wild-type phenotype in all the assays used in this study.
AGS SIB02 cells expressing p65-GFP were infected with the parental strain WT DIE2 or with the corresponding mutants: DescV, DnleE, DnleBE, and DnleE complemented with a plasmid expressing nleE. After 3 h, cells were TNFa-induced for 30 min, stained with Hoechst 33342 and analyzed by automated microscopy. The results show that whereas the wild type repressed ,90% of the p65 translocation to the nucleus, the TTSS escV mutant was attenuated, exhibiting only ,50% repression (Fig. 4A). These results indicate that EPEC inhibit p65 translocation by both TTSS-independent and TTSSdependent pathways. Importantly, the nleE and nleBE mutants were as deficient as the escV mutant in blocking p65 translocation. Moreover, complementation of the nleE mutant with the wild-type nleE allele restored the bacteria's full capacity to inhibit the p65 translocation (Fig. 4A). These results support the hypothesis that the TTSS-mediated inhibition of p65 translocation is NleE-dependent. In addition, our results suggest that a putative TTSS-independent mechanism might function in parallel to NleE to inhibit p65 translocation.

NleE is required for full inhibition of TNFa-induced IL-8 expression
To further substantiate our results, we used IL-8 expression as an additional read-out for NF-kB activation. Briefly, HeLa cells were infected with different EPEC strains or remained uninfected.
Then, cells were washed and treated for 3 h with TNFa and gentamycin, to kill the remaining bacteria. RNA was then extracted from the cells and the amount of produced IL-8 mRNA was measured by real time PCR. In comparison to non infected cells or cells infected with EPEC escV mutant, both wild type and the DIE2 mutant exhibit a ,100 fold repression of IL8 expression (Fig. 4B). In contrast, the nleE mutant exhibited a partial, less then 10 fold, repression of IL8 expression and this was moderately complemented by plasmid expressing native NleE (Fig. 4B). A more severe deficiency in repression of IL8 expression was exhibited by a double mutant nleBE (Fig. S3). Furthermore, a plasmid expressing nleBE restored IL8 repression to that seen in wild type EPEC (Fig. S3). Upon testing the amount of secreted IL8 protein instead of production of IL8 mRNA, similar results were obtained (Fig. S2). Taken together these results show that (i) NleE is required for full inhibition of IL-8 expression, (ii) NleB also contributes to this repression and iii) a putative TTSS effector(s), other then NleB and NleE might function in parallel to inhibit IL-8 expression.  NleE is required to inhibit the EPEC-induced IL8 expression The IL8 expression assay was found to be much more sensitive then testing translocation to the nucleus or the IkB degradation assay. This is probably since the latter are very transient events while the mRNA tends to accumulate, increasing the signal/noise ratio. Interestingly, using the IL8 expression assay we found that infection with the escV mutant was sufficient to induce IL8 expression in HeLa cells, albeit not as strong as that induced by TNFa (data not shown). This activation is possibly via the activity of flagellin, LPS or other PAMPs. We thus next asked whether NleE also inhibits the EPEC-induced IL8 expression. To this end we repeated the experiment described in Fig. 4B, but TNFa was omitted. We found that even the non infected cells produce certain levels of IL8 mRNA, but upon infection with EPEC escV mutant we observed a ,10 fold increase in IL8 expression (Fig. 4C). In contrast, the EPEC wild type (or the DIE2 mutant) exhibited strong repression of the EPEC-induced IL8 expression. Importantly, the nleE mutant exhibited only a partial capacity to repress the self-induced IL8 expression. Similar results where observed when we used the double mutant nleBE instead of nleE mutant (Fig.  S3). However, both the nleE or the nleBE, mutants were not as deficient in IL8 repression as the escV mutant ( Fig. 4C and S3). Thus, we predict that additional putative effector might function in parallel to NleB and NleE to repress IL8 expression. In conclusion, our results clearly show that i) EPEC mediate a TTSSdependent repression of self-induced IL8 expression; and ii) NleE is required for full repression of the EPEC-induced IL8 expression.

NleE, expressed by HeLa cells, inhibits NF-kB translocation to the nucleus
We next examined whether NleE is sufficient for inhibition of NF-kB activation in the absence of the infecting bacteria and other putative effectors. To this end, we constructed a vector expressing mCherry fused to NleE (mCherry-NleE) and used it for transient transfection of HeLa cells. Untransfected cells (Fig. 5A) or cells transfected with either the mCherry-NleE vector or a vector expressing mCherry alone (Fig. 5B) were stimulated with TNFa for 1 h, or remained untreated. Next, these cells were fixed, stained with anti-p65 antibody, and analyzed by fluorescent microscopy to determine both the ability of the transiently expressed NleE to inhibit TNFa-induced migration of p65 to the nucleus and to determine its localization in the expressing cells. The TNF treatment induced strong migration of p65 to the nucleus in the untransfected cells (Fig. 5A), and in cells transfected with the negative control vector (Fig. 5B two upper panels and 5C). Importantly, the transiently expressed mCherry-NleE induced a strong inhibition of p65 translocation to the nucleus (Fig. 5B two lower panels and 5C). The expressed mCherry and mCherry-NleE were similarly distributed in the cells, predominantly in the cytoplasm (Fig. 5B). These results indicate that NleE IE6 is sufficient for inhibition of NF-kB migration to the nucleus presumably by IkB stabilization. Similar analysis using NleE IE2 instead of NleE IE6 , show that NleE IE2 lost the ability to block p65 translocation to the nucleus (Fig. S4), highlighting the importance for NleE activity of the region between residues 49-115, which is deleted in NleE IE2 (Fig. S1).

NleE inhibits phosphorylation of IkB
Different NF-kB activating pathways converge at the level of IKK phosphorylation, which subsequently leads to IkB phosphorylation, targeting it to ubiquitination and proteasome-mediated degradation [6]. We thus tested whether NleE inhibits the TNFainduced IkB phosphorylation. Cells were infected with different EPEC strains followed by TNFa treatment. The levels of IkB and phospho-IkB were then determined by immunoblot analysis with the appropriate antibodies and the relative accumulation of unphosphorylated IkB was determined. For a negative control, we used cells infected with the escN mutant, which cannot stabilize IkB (Fig. 1A). Indeed, in cells infected with this mutant we noted increased IkB phosphorylation followed by its degradation. However, the addition of proteasome inhibitor (MG132) resulted in accumulation of phosphorylated IkB (Fig. 6A). As a parental strain, we used the DIE2 strain (WT DIE2 ), which, like wild-type EPEC, efficiently protected IkB from degradation ( Fig. 1A and 6A). Importantly, the accumulated IkB in these cells was mostly unphosphorylated. In contrast, the corresponding nleE mutant failed to induce accumulation of unphosphorylated IkB, exhibiting a phenotype similar to that of the escN mutant (Fig. 6A). Taken together, these results indicate that wild-type EPEC stabilizes IkB by preventing its phosphorylation and that NleE is required for this activity. Indeed, complementing the DnleE mutant with a plasmid expressing NleE restored the bacteria's capacity to induce the accumulation of unphosphorylated IkB (Fig. 6A).
These results suggest that one can restore the inability of the DnleE mutant to prevent IkB degradation by two alternative approaches: (i) by treatment with proteosome inhibition, to inhibit phospho-IkB degradation, or (ii) by complementation with a plasmid expressing nleE, to block IkB phosphorylation. To compare the efficiency of these two treatments, we infected HeLa cells with the DIE2 strain (WT DIE2 ), DnleE mutant, or with the DnleE mutant complemented either by proteosome inhibitor (MG132) treatment upon TNFa induction, or by a plasmid expressing nleE. The results show that both treatments similarly stabilized the IkB. However, the first treatment led to a strong phosphorylation of the accumulated IkB whereas when NleE was added, the accumulated IkB remained unphosphorylated (Fig. 6B). These results further support the notion that NleE stabilizes IkB by inhibiting its phosphorylation.

NleE inhibits IKKb activation induced by TNFa or IL1b
The signaling pathways induced by the TNF receptor (TNFR) is different from that induced by the IL1 or TLR receptors, but both converge at the level of IKK activation by TAK1 (Fig. 6C, [16]). The inhibition of the self-induced IL8 expression by NleE (Fig. 4C), is hinting that NleE functions downstream to the pathways converging point. To directly test this prediction we tested whether EPEC is capable of inhibiting IL1b-induced degradation of IkB. Importantly, we found that wild type EPEC, but not the nleE mutant, inhibited the IL1b-induced IkB degradation (Fig. 6D). These results confirmed that NleE functions downstream to the signaling converging point. We next tested whether NleE can block the phosphorylation and thus activation of IKKb. To this end we extracted proteins from cells, which were infected with different strains and then treated with TNFa or IL1b as indicated (Fig. 6E). The extracted proteins were subjected to western analysis using anti-IKKb, anti-phospho-IKK, anti-IkB and anti- phospho-IkB antibodies. The results show that, treatment with either TNFa or IL1b induced IkB and IKK phosphorylation in non infected cells or cells infected with the escV mutant. We also found that wild type EPEC, but not the nleE mutant, inhibited this IKK phosphorylation. The same inhibition is noted for the IkB phosphorylation, in the wildtype strain, However, due to IkB degradation, less protein is noted and thus its phosphorylation cannot be seen (Fig. 6E). Complementation with plasmid expressing wild type nleE allele only partially, but consistently restored the inhibition of IKKb phosphorylation (Fig. 6E). These results suggest that NleE block activation of IKKb. Taken together our results indicate that NleE blocks the NF-kB signaling cascade downstream to the converging point of the TNFa and IL1b signaling pathways, but upstream to IkB phosphorylation, possibly by direct blocking TAK1 or IKKb activation.

Discussion
In this report we showed that NleE of EPEC stabilizes the NF-kB inhibitor, IkB, via inhibition of its phosphorylation, thereby preventing NF-kB signaling. This activity of NleE was discovered using an unbiased screen of EPEC strains deleted of most of the EPEC-specific genes. We showed that an nleE mutant is deficient in blocking IkB phosphorylation and in preventing its degradation. Moreover, an nleE mutant was attenuated in blocking TNFainduced NF-kB migration to the nucleus as well as in IL-8 expression and secretion. These abilities were restored to the mutant upon complementation with a plasmid expressing the wildtype nleE allele. Importantly, we showed that NleE expressed in HeLa cells blocks NF-kB translocation to the nucleus upon TNFa treatment. Taken together, our findings indicate that NleE is sufficient to inhibit NF-kB signaling by blocking IkB phosphorylation. Further analysis suggest that NleE blocks the NF-kB signaling cascade downstream to the converging point of the TNFa and IL1b signaling pathways, but upstream to IkB phosphorylation, possibly by directly blocking TAK1 or IKKb activation.
We also show that NleB enhances NleE activity. The nleE mutant still showed residual inhibition of IkB degradation, which was eliminated upon further deletion of nleB. Moreover, complementation of the nleBE double mutant with a plasmid expressing nleBE was more efficient than a plasmid expressing only nleE. These results suggest that NleB plays a role in IkB stabilization. Similar results were obtained when expression of IL8 was used as a readout for inhibition of the NF-kB signaling. The mechanism underlying NleB function is not yet apparent. Nevertheless, the notion that NleB and NleE function together is supported by the facts that nleE form a putative bicistronic operon with nleB and that nleE is consistently associated with nleB in natural isolates of diarrheagenic EPEC [17]. Other isolates of EPEC, EHEC, and C. rodentium carry nleB and nleE alleles almost identical to nleBE IE6 investigated in this study (Supplemental material Fig. S1). We predict that all NleBE proteins function similarly.
The phenomenon of effectors functioning in parallel is common in EPEC [1]. Interestingly, the TTSS mutants (escV) and DnleBE were only partially deficient in blocking the TNFa-induced migration of NF-kB to the nucleus, suggesting that an additional TTSS-independent mechanism might function in parallel to NleBE to inhibit translocation of NF-kB to the nucleus (Fig. 7). This activity might be related to damping of the signal due to continuous exposure to PAMPs like flagellin or LPS. Interestingly, the TTSS escV mutant was completely deficient in inhibition of IL8 expression, while the nleBE mutant was only partially deficient in this function. We therefore predict that EPEC encode additional putative effector(s) that function in parallel to NleBE by blocking IL-8 expression (Fig. 7).
Where as in this study we showed that NleE represses NF-kB activation, a recent report suggested exactly the opposite, i.e. NleE and OspZ, an NleE-homolog encoded by Shigella, activates NF-kB signaling [18]. The discrepancy between the two studies might result from the use of different cell lines. In their study, Zurawski et al. [18] used cell lines that activate NF-kB upon sensing of either TNFa or bacterial PAMPs including LPS. The latter might complicate the interpretation of the results. To avoid such complications, we used in this study HeLa and AGS cells, which are inert to LPS. Moreover, we adjusted the experimental conditions used in this study, so that the NF-kB activation by other bacterial PAMPs was insignificant. This facilitated the uncoupling of effector injection and NF-kB activation, which occurred only upon addition of TNFa. An alternative, although less likely explanation for the inconsistency between the two studies is that NleE functions differently in different types of cell or different cell lines.
Like EPEC, Yersinia also employs an injected effector, YopJ, to block IkB phosphorylation. YopJ is an acetyl transferase that acetylates critical IKKb residues and thus prevents its activation [19]. Although both NleE and YopJ block IKKb phosphorylation, they are very different in sequence, which probably reflects functional differences. We are currently investigating the NleE's mode of action. Other effectors that interfere with NF-kB function include the Salmonella SspH, and Shigella IpaH9.8, which are targeted to the host cell nucleus and inhibit NF-kB-dependent transcription [20]. Shigella also uses OspG, which inhibits ubiquitination of phospho-IkB [15]. Interestingly, EPEC encode an OspG homolog, NleH. However, we found that an EPEC strain, mutated in its two nleH alleles, still inhibit IkB degradation. Yet, our analysis indicates that additional TTSS effector(s) inhibits NF-kB signaling (Fig. 7) and we can not exclude the possibility that this putative effector is nleH. Signals (indicated as lightning) sensed by different receptors elicit signal transduction cascades that converge at the level of TAK1 activation and IKK phosphorylation, which subsequently leads to IkB phosphorylation. This step is inhibited by NleBE. NleE inhibits IKKb phosphorylation, but the exact target of NleE and NleB is yet to be defined. The phosphorylated IkB is subjected to ubiquitination and proteasomemediated degradation, allowing NF-kB translocation to the nucleus and activation of expression of target genes including IL-8. Our results predicate that in addition to NleBE, a putative non-TTSS factor and a putative TTSS effector might inhibit NF-kB translocation to the nucleus and IL-8 expression, respectively (Fig. 4). doi:10.1371/journal.ppat.1000743.g007 The role of NleE in EPEC virulence was not tested, since an animal model is not yet available, but it was tested with C. rodentium. Importantly, NleE was found to be required for full virulence of C. rodentium upon infection of wild-type mice, but this requirement was diminished upon infection of mice deficient in TLR4 [21,22]. Our results revealed the rationale behind this intriguing phenomenon. We argue that NleE-mediated NF-kB repression is no longer needed if the host itself is deficient in TLR4/LPS-induced NF-kB signaling.
In conclusion, we show that NleE blocks IkB phosphorylation by IKK and thus it inhibits NF-kB signaling. We also show that NleB enhances NleE's activity and that EPEC probably use additional mechanisms to interfere with other constituents of the NF-kB signaling pathway. This is presumed to have multiple consequences on the course of EPEC infection and the maturation of both innate and adaptive host immune response.

Bacterial strains, plasmids, and primers
Bacterial strains, plasmids, and primers used in this study are listed in Table S1, S2, and S3, respectively (Supplemental material). Deletions in the EPEC chromosome were constructed using the primers listed in Table S3, as described [23]. For bacterial expression, the genes were cloned in the pSA10 expression vector as described [24]. In most cases the leakiness of the Tac promotor was enough for gene expression. When indicated, IPTG (0.01mM) was added. For the cloning procedure, genes were amplified by RCR using the primers listed in Table S3. Formation of plasmid-borne nleE-blaM fusions were carried out as described [25] using the plasmid pCX341 and primers as listed in Table S3. The nleE IE2 DNA was amplified from EPEC DIE6 mutant and that of nleE IE6 from EPEC DIE2 mutant. For formation of mCherry fusions, the EGFP gene of pEGFP-N1 (Clonetech) was excised from the plasmid using the NotI and BamHI and replaced by mCherry taken from pREST-mCherry, resulting in pMS2841. This plasmid was transformed into pSC4141 by introduction of a unique scaI site at the mCherry 39, eliminating its stop codon in the process. This was done using QuikChange site-directed mutagenesis kit (Stratagene #200518-5) and primers listed at Table S3. This plasmid was further used to create the transcriptional fusion: mCherry-nleE IE6 -6his (pSC4144) and mCherry-nleE IE2 -6his (pSC4350), using primers listed at Table  S3.

Analysis of IkB stability and IkB and IKKb phosphorylation
HeLa cells (9610 5 ) in 4 cm plates were infected with a 1:100 dilution in DMEM of bacteria grown overnight statically at 37uC (multiplicity of infection, MOI,1:100). Following 3 h infection in 5% CO2, at 37uC, the medium was replaced with fresh DMEM with or without either 10 ng/ml TNFa for 40 min (or 20 ng/ml IL-1b for 20 mins). When indicated, the infecting bacteria were supplemented with 0.01 mM IPTG at 1.5 h post inoculation. To terminate the infection, cells were washed with 3 ml of cold TBS (20 mM Tris-HCl, pH 7.4, 150 mM NaCl), scraped with 1 ml of cold TBS, collected and centrifuged, (800g, 2 min at 4uC). The pellet was resuspended in 40 ml lysis buffer (0.5% Triton-6100, 20mM Tris-HCl pH 7.2, 0.2 mM VO 4 , 10 mM NaF, 30 ml of complete inhibitor Roche) and centrifuged (20,000g, 3 min at 4uC). Supernatant was either transferred for protein quantification assay (BCA assay) or to a tube with loading dye (LDS sample buffer, NuPAGE), boiled for 10 min, and then centrifuged (20,000g, 3 min). Samples were quantified using bicinchoninic acid (BCA) and copper sulfate. Equal protein concentration for each sample was then loaded on SDS-PAGE gel, transferred to PVDF membrane, and reacted with antibodies against IkB (1:1000), Tubulin, as a loading verification control (1:2500), or phospho-IkB (1:1000, Cell Signaling). When indicated, 20 mM MG132 (1:1000) was used. Protein band density was quantified using Tina software (version 2.09) and the percentage of the unphosphorylated IkB was determined by calculating the relative phosphorylated IkB out of the total IkB shown for each lane. IKKb analysis was done as described for IkB except that induction time with TNFa was reduced to 10 min and with IL1b it remained 20 min. IKK detection was preformed by western blot analysis using anti-IKKb antibody (1:1000, Cell Signaling Technology, #2684) and Phospho-IKKa (Ser180)/IKKb (Ser181) antibody (1:1000, Cell Signaling Technology, #2681S)

Analysis of p65 translocation to the nuclei
Generation of the p65-GFP-expressing cell line and the automated image analysis to quantify translocation of p65-GFP is described elsewhere (Bartfeld et al., submitted). Briefly, SIB02 cells are AGS cells lentivirally transduced to express p65-GFP. SIB02 cells, seeded in 96-well-plates, were inoculated with EPEC strains at MOI 1:100, incubated for 3 h and subsequently activated by 10 ng/ml TNFa. After an appropriate incubation time, cells were fixed using 100% ice-cold methanol and stained with Hoechst 33342 (2 mg/ml). Images of ,200 cells were acquired using automated microscopy (Scan ' R, Olympus) and translocation of p65-GFP to the nucleus was subsequently quantified using Scan ' R image analysis software (Olympus) as described (Bartfeld et. al., submitted). Cells with nuclear p65-GFP above the defined threshold were termed ''active'' and the percentage of active cells per well was calculated.

Expression and translocation of NleE-BlaM fusions
To determine translocation levels, overnight cultures of wildtype EPEC containing plasmids expressing NleE-BlaM were diluted 1:50 in DMEM and used to infect HeLa cells for 3 h. Cells were then washed and stained with CCF2 for 2.5 h as described [26], washed in cDMEM, excited at 405 nm, and then emission at 465 nm and 535 nm was recorded (SPECTRAFluor, TECAN). The amount of translocation was determined as described [26]. As a negative control, we used EPEC expressing unfused BlaM (Vector). To determine expression levels, the unattached bacteria were harvested, washed, and lysed by repeated freezing and thawing in PBS containing 1 mM EDTA, 1 mg/ml lysozyme, and 0.1% Triton-6100. The BlaM activity in the lysate was determined using nitrocefin as substrate and the rate of product accumulation per number of bacteria (OD 600) was determined as described [25].

IL-8 expression
HeLa cells (7610 5 ) in 6 wells plates were inoculated with a 1:100 dilution in DMEM of bacteria grown overnight statically at 37uC (multiplicity of infection, MOI,1:100) and incubated for 3 h (5% CO 2 , 37uC). To terminate the infection and induce IL8 expression, the medium was replaced with fresh DMEM supplemented with 2% FCS, 100ug/ul gentamicin and with or without 10 ng/ml TNFa and incubated for additional 3 h. Cells were than washed with 2 ml of cold TBS (20 mM Tris-HCl, pH 7.4, 150 mM NaCl), scraped with 1 ml of cold TBS, collected and centrifuged, (800 g, 2 min, 4uC). RNA was extracted using the MasterPure Complete DNA and RNA Purification Kit (EPICEN-TRE Biotechnologies) and used to synthesize cDNA with the Verso cDNA kit (Thermo scientific). hHPRT transcript levels were used to normalize total RNA levels in samples. Real time analysis was than conducted using Absolute Blue QPCR SYBR Green (Thermo scientific) in a real-time cycler (Rotor-Gene 6000, Corbett).

Supporting Information
Found at: doi:10.1371/journal.ppat.1000743.s004 (1.58 MB TIF) Figure S2 NleE is involved in blocking IL8 secretion. HeLa cells (8610 4 per well, seeded in 24-well plate) were infected for 3h with different EPEC strains as indicated or remained uninfected (N/I). After 3.5 h, supernatants were replaced with 300 ul DMEM, 2% FCS, and 50 mg/ml gentamycin with or without 10 ng/ml TNFa. After 16 h, 100 ml of cleared supernatant taken from each well was used for IL-8 measurements using Human CXCL8/IL-8 Quantikine immunoassay assay (R & D), according to the manufacturer's instructions. PBS and IL-8 were used as negative and positive controls for the detection assay. The relative amounts of IL-8 are shown. The experiment was done twice in duplicates and typical results are shown. Standard errors are indicated by bars. In the case of DescV (indicated by a vanishing colored bar), the signal was above the upper limit of the detection levels. Untreated and uninfected cells secreted ,300 units of IL-8, whereas TNFa treatment induced a 10-fold increase in IL-8 secretion (,3000 units) (Fig. 4B). In contrast, pre-infection with wild-type EPEC or with the DIE2 mutant (WT DIE2 ), reduced IL-8 secretion below the detection levels, even upon TNFa treatment. Furthermore, the TTSS escV mutant was completely deficient in blocking IL-8 secretion. In conclusion, EPEC strongly reduces IL-8 secretion by a TTSS-dependent mechanism. Importantly, we found that the DnleE or DnleEB mutants were strongly deficient in blocking IL-8 secretion but not as deficient as the TTSS mutant (escV). These results show that (i) NleE is required for full inhibition of IL-8 secretion and (ii) other putative TTSS effector(s) might function in parallel to NleE to inhibit IL-8 expression and/or secretion. Found at: doi:10.1371/journal.ppat.1000743.s005 (1.62 MB TIF) Figure S3 A mutant baring a deletion of nleBE is deficient in repressing self-activated, or TNFa-induced, IL8 repression. HeLa cells were infected with the indicated strains and analyzed as described in Fig. 4B and 4C. To terminate the infection and induce IL8 expression, the medium was replaced with fresh DMEM supplemented with 2% FCS, 100ug/ul gentamicin and with or without 10 ng/ml TNFa and incubated for additional 3 h. Cells were than washed with 2 ml of cold TBS (20 mM Tris-HCl, pH 7.4, 150 mM NaCl), scraped with 1 ml of cold TBS, collected and centrifuged, (800 g, 2 min, 4uC). RNA was extracted using the MasterPure Complete DNA and RNA Purification Kit (EPICEN-TRE Biotechnologies) and used to synthesize cDNA with the Verso cDNA kit (Thermo Scientific). hHPRT transcript levels were used to normalize total RNA levels in samples. Real time analysis was than conducted using Absolute Blue QPCR SYBR Green (Thermo Scientific) in a real-time cycler (Rotor-Gene 6000, Corbett). The amount of IL-8 mRNA in each strain is shown as a percentage of the level relatively to the transcript levels in the DescV mutant. The experiment was done twice in duplicates and typical results are shown. Bars indicate standard errors. Found at: doi:10.1371/journal.ppat.1000743.s006 (0.44 MB TIF) Figure S4 NleE IE2 is deficient in blocking TNF-induced translocation of p65 to the nucleus. HeLa cells were transected with plasmids expressing mCherry, mCherry-NleE IE2 or mCherry-NleE IE6 . The expressing cells were treated with TNFa for 1 h, or remained untreated, after which they were fixed and stained with anti-p65. The slides were analyzed by fluorescent microscopy and the percentage of red cells (expressing mCherry or mCherry fused to NleE IE6 or NleE IE2 ) containing nuclear p65 was determined. The number of cells quantified is indicated and standard errors are indicated by bars. The results show that while NleE IE6 inhibited p65 translocation, NleE IE2 failed to do so.