Salmonella enteritidis Effector AvrA Stabilizes Intestinal Tight Junctions via the JNK Pathway*

Salmonella pathogenesis studies to date have focused on Salmonella typhimurium, and the pathogenesis of a second major serotype, Salmonella enteritidis, is poorly understood. Salmonella spp. possess effector proteins that display biochemical activities and modulate host functions. Here, we generated a deletion mutant of the effector AvrA, S.E-AvrA−, and a plasmid-mediated complementary strain, S.E-AvrA−/pAvrA+ (S.E-AvrA+), in S. Enteritidis. Using in vitro and in vivo infection models, we showed that AvrA stabilizes epithelial tight junction (TJ) proteins, such as ZO-1, in human intestinal epithelial cells. Transepithelial electrical resistance was significantly higher in cells infected with S.E-AvrA+ than in cells infected with S.E-AvrA−. Inhibition of the JNK pathway suppresses the disassembly of TJ proteins; we found that enteritidis AvrA inhibited JNK activity in cells infected with wild type or S.E-AvrA+ strains. Therefore, Enteritidis AvrA-induced ZO-1 stability is achieved via suppression of the JNK pathway. Furthermore, the S.E-AvrA− strain led to enhanced bacterial invasion, both in vitro and in vivo. Taken together, our data reveal a novel role for AvrA in S. Enteritidis: Enteritidis AvrA stabilizes intestinal TJs and attenuates bacterial invasion. The manipulation of JNK activity and TJs in microbial-epithelial interactions may be a novel therapeutic approach for the treatment of infectious diseases.

Salmonella spp. are Gram-negative, facultative anaerobes, and intracellular pathogens in both humans and animals that pose a major public health problem worldwide. It is estimated that more than 1 million people acquire a Salmonella infection as a foodborne illness and that such infections contribute to Ͼ350 deaths annually in the United States (1) and 93.8 million illnesses and 155,000 deaths worldwide each year (2). In the past 20 years, Salmonella enterica serotype enteritidis (S. Enteritidis) 3 has become one of the most common serotypes reported in human cases of salmonellosis worldwide, despite ongoingimplementationoftargetedcontrolandpreventionmeasures (3)(4)(5). In humans, S. Enteritidis salmonellosis causes symptoms such as abdominal pain, diarrhea, nausea, vomiting, fever, and headache (6). S. Enteritidis is also increasingly reported in cases of invasive and extraintestinal infections, such as septicemia, arthritis, endocarditis, meningitis, and urinary tract infections (7)(8)(9)(10)(11)(12)(13). Most Salmonella pathogenesis studies to date have focused on Salmonella typhimurium; thus, the pathogenesis of another major serotype, Salmonella Enteritidis, is poorly understood.
Salmonella species possess a range of effector proteins that are translocated into host cells via a type III secretion system (14,15). These bacterial effector proteins display a large repertoire of biochemical activities and modulate the function of host regulatory molecules, including tight junction (TJ) proteins and their upstream regulators (16 -20). AvrA is a Salmonella effector protein secreted by the Salmonella pathogenicity island 1 (SPI-1) type III secretion system. The AvrA protein in S. typhimurium is an anti-inflammatory effector that processes acetyltransferase activity toward specific host MAPKKs and inhibits the host JNK/AP-1 and NF-B signaling pathways (21)(22)(23)(24)(25).
The intestinal epithelium is formed by a single layer of epithelial cells that separates the intestinal lumen from the underlying lamina propria. The space between these cells is sealed by TJs, which regulate the permeability of the intestinal barrier (26). TJs are the most apical component of intercellular junctional complexes formed by integral membrane proteins such as occludin, claudin family members, and junctional adhesion molecules. The cytoplasmic domain of occludin and claudins binds to peripheral membrane proteins known as the zonula occludens (ZO), such as ZO-1, ZO-2, and ZO-3 (27,28), which separate the apical and basolateral cell surfaces to provide a barrier function, inhibit solute and water flow through the paracellular space, and inhibit bacterial invasion (29). Signaling pathways involved in the assembly, disassembly, and maintenance of TJs are controlled by numerous signaling pathways, such as the protein kinase C, MAPK, myosin light chain kinase, and Rho GTPase pathways (30 -33).
The JNKs were originally identified based on their activation in response to environmental stress stimuli, such as bacterial infection. JNK binds and phosphorylates c-Jun within its transcriptional activation domain, and this signaling pathway generally contributes to inflammatory responses (34). The JNK pathway plays a role in regulating TJ integrity in different epithelial cells (25,35,36). Previous studies on AvrA were based on S. typhimurium and were performed using in vitro and mouse models (37). However, the role of the bacterial effector AvrA in S. Enteritidis remains unknown. We hypothesized that S. Enteritidis AvrA stabilizes intestinal epithelial TJs and decreases bacterial invasion. Using S. Enteritidis-infected intestinal epithelial cells and a mouse colitis model, we found that enteritidis AvrA may reprogram the regulation of TJs in eukaryotic cells to benefit the pathogen.

Establishment of an AvrA Deletion Mutant and a Plasmidmediated Complemented Strain in S. Enteritidis-We first established an
AvrA deletion mutant, S.E.-AvrA Ϫ , in S. Enteritidis strain C50336 using -Red-mediated recombination and complemented by plasmid pBR322-AvrA (Fig. 1). Our PCR data showed that the AvrA gene was deleted in the S.E-AvrA Ϫ mutant and restored in the complemented strain S.E-AvrA ϩ (Fig. 1A). RT-PCR verification of the AvrA gene transcription further confirmed the successful generation of the AvrA mutant and the complemented strain, with the GMK bacterial reference gene for normalization (Fig. 1B). Our Western blot data showed that the AvrA protein was absent in the mutant strain (S.E-AvrA Ϫ ) and overexpressed in the complemented strain (S.E-AvrA ϩ ) (Fig. 1C). Ponceau S staining indicated even protein loading in each lane. The growth characteristics of the strains were determined in LB liquid medium; there was no difference between the AvrA mutant or complemented strains and the wild type C50336 strain (Fig. 1D). The mutant and complemented strains were characterized biochemically using the API 20E identification kit, which revealed no differences between the mutant and complemented strains relative to the wild type strain (Fig. 1E).
Enteritidis AvrA Stabilizes Epithelial TJ Proteins and Alters Transepithelial Electrical Resistance in Vitro-ZO-1 is a TJ protein on the surface of intestinal epithelial cells. In human intestinal epithelial Caco-2 BBE cells, ZO-1 protein levels were decreased after S.E-AvrA Ϫ infection compared with wild type S.E or S.E-AvrA ϩ infection ( Fig. 2A). Densitometry analysis showed a significant difference between cells infected with S.E-AvrA Ϫ and those infected with either wild type S.E or S.E-AvrA ϩ (Fig. 2, B and C). In contrast, there were no significant changes in the TJ proteins occludin, claudin-1, and claudin-7 or the adhesion junction protein E-cadherin with or without S. Enteritidis colonization ( Fig. 2A). We observed similar trends for ZO-1 following S.E-AvrA Ϫ , S.E-AvrA ϩ , or wild type S.E infection in HCT116 and SKCO15 cells (Fig. 2, D and E). RT-PCR data showed no change in ZO-1 mRNA expression (Fig. 2,

F-H).
We further determined the distribution of ZO-1 protein in Caco-2 BBE cells by immunofluorescence. As shown in Fig. 3, ZO-1 localization was disrupted in S.E-AvrA Ϫ -infected cells but was unchanged in S.E-AvrA ϩ or wild type S.E-infected cells. There was no change in the adhesion junction protein E-cadherin with or without S. Enteritidis colonization (see Fig. 6E).
The disruption of TJs results in greater intestinal epithelial cell permeability (38). We next measured the transepithelial electrical resistance (TEER) in human epithelial Caco-2 BBE cells (Fig. 3B). There was a dramatic decrease in TEER 1-8 h after S. Enteritidis infection, and this decrease was greater in S.E-AvrA Ϫ -infected cells than in S.E-AvrA ϩ or wild type S.Einfected cells (Fig. 3B). There was a significant difference between S.E-AvrA Ϫ -infected cells and wild type S.E or S.E-AvrA ϩ -infected cells 2-8 h after S. Enteritidis infection (Fig.  3B).
Enteritidis AvrA Inhibits Cellular JNK Activation-ZO-1 is regulated by the JNK signaling pathway (25,39,40). Using Western blotting, we found that JNK signaling was activated, as indicated by the marked increase in JNK phosphorylation, in Caco-2 BBE, HCT116, and SKCO15 cells following S. Enteritidis colonization (Fig. 2). There was greater JNK phosphorylation in S.E-AvrA Ϫ -infected cells than in wild type S.E or S.E-AvrA ϩ -infected cells. JNK pathway activation induces IL-8 production in epithelial cells (41), and JNK pathway-dependent expression of this pro-inflammatory cytokine is shown in Fig. 4. In parallel with p-JNK, IL-8 levels in culture supernatant were higher 6 h after colonization with S.E-AvrA Ϫ compared with wild type S.E and S.E-AvrA ϩ . Together, these results show that AvrA inhibits JNK activation in epithelial cells upon infection by S. Enteritidis. ZO-1 protein levels were inversely correlated with phosphorylated JNK levels. In S.E-AvrA Ϫ -infected cells, increased JNK phosphorylation correlated with reduced ZO-1 expression, whereas in wild type S.E or S.E-AvrA ϩ -infected cells, ZO-1 protein levels did not change.
Enteritidis AvrA Deletion Leads to Enhanced Bacterial Invasion-The disruption of TJs results in greater intestinal epithelial cell permeability, as shown by the TEER changes in the epithelial cells following colonization by S. Enteritidis. Next, we determined the efficiency of bacterial invasion in human epithelial Caco-2 BBE cells. As shown in Fig. 4B, the S.E-AvrA Ϫ -colonized cells showed an increased intracellular bacterial load compared with those colonized with wild type S.E or S.E-AvrA ϩ . Therefore, we conclude that disrupting TJs also results in enhanced invasion of epithelial cells by S. Enteritidis.
Enteritidis AvrA Deletion Leads to Enhanced Pathological Lesions in the Mouse Cecum-To study the role of the S. Enteritidis protein AvrA in an in vivo model of natural intestinal infection, we used the streptomycin pretreatment mouse model of enteric salmonellosis (42,43). 6 -8-week-old female C57BL/6 mice were pretreated with streptomycin for 24 h before infection with wild type S. Enteritidis C50336, S.E-AvrA Ϫ , or S.E-AvrA ϩ by oral gavage.
Mice cecal histopathology was graded with a semiquantitative scale (43); morphological changes were observed following S. Enteritidis infection, including goblet cell loss, polymorphonuclear cell infiltration into the lamina propria, and epithelial damage (Fig. 5A). Mice infected with S.E-AvrA Ϫ showed a marked increase in the pro-inflammatory response compared with mice infected with the wild type S.E or S.E-AvrA ϩ strain (Fig. 5, B and C).
Enteritidis AvrA Stabilizes the TJ Protein ZO-1 and Inhibits the JNK Pathway in the Mouse Colon-In vitro, we found that the S. Enteritidis protein AvrA inhibits the JNK pathway and stabilizes the TJ protein ZO-1. In colonic samples from S.E-AvrA Ϫ -infected mice, ZO-1 protein expression was signifi-cantly decreased (Fig. 6, A and B) in parallel with increased JNK phosphorylation compared with samples from wild type S.E or S.E-AvrA ϩ -infected mice (Fig. 6, A and C). No changes were observed in the expression levels of claudin-7, occluding, or E-cadherin (Fig. 6A). RT-PCR showed that ZO-1 mRNA levels in the mouse colon were not altered by S.E-AvrA Ϫ , S.E-AvrA ϩ or wild type S.E infection (Fig. 6D). Immunofluorescence analysis of mouse colonic tissues further confirmed that ZO-1 localization was disrupted in S.E-AvrA Ϫ -infected mice compared with S.E-AvrA ϩ -or wild type S.E-infected mice (Fig. 6E). Similar to the Western blotting results, there was no change in claudin-7 or E-cadherin (Fig. 6E) by immunofluorescence.

AvrA Deletion Leads to Increased Bacterial Invasion and
Inflammation in the Mouse Colon-Next, we investigated bacterial invasion in the mouse colon. In our model, bacterial colonization was measureable in the mouse colon. There was a significantly higher bacterial load in S.E-AvrA Ϫ -colonized colon tissue than in wild type S.E or S.E-AvrA ϩ -colonized colon tissue (Fig. 7A), whereas there was no evidence of Salmonella in the uninfected control mice.
We also measured serum levels of pro-inflammatory cytokines in S. Enteritidis-infected mice, using Luminex assays (Fig.  7, B and C). The pro-inflammatory cytokines IL-6 ( Fig. 7B) and TNF-␣ (Fig. 7C) were not increased at 8 h after Salmonella infection. However, after 4 days, the concentrations of IL-6 and TNF-␣ were markedly increased in S.E-AvrA Ϫ -infected mice compared with wild type S.E or S.E-AvrA ϩ -infected mice. We confirmed the changes in pro-inflammatory cytokine expression at the mRNA level using RT-PCR: IL-6 ( Fig. 7D) and TNF-␣ (Fig. 7E). mRNA expression was significantly increased in S.E.-AvrA Ϫ -infected colons. Our results indicate that S. Enteritidis AvrA decreases bacterial invasion and reduces inflammation.
Inhibition of the JNK Pathway Abolishes the Effect of AvrA Deletion on ZO-1-To confirm whether AvrA stabilizes the TJ protein ZO-1 via the JNK pathway, we treated Caco-2 BBE cells with a JNK inhibitor (SP600125) to determine whether inhibiting the JNK pathway destabilizes ZO-1 in S.E-AvrA Ϫ -infected cells. As shown in Fig. 8 (A-C), after treatment with the JNK inhibitor, phosphorylated JNK was nearly undetectable in S.E-AvrA Ϫ , S.E-AvrA ϩ , or wild type S.E-infected cells. However, the JNK inhibitor did not disrupt ZO-1 expression in S.E-AvrA Ϫinfected cells. RT-PCR supported the observation that the JNK inhibitor had no effect on ZO-1 mRNA expression (Fig. 8D). Inhibiting the JNK pathway decreased the expression of the JNK-dependent pro-inflammatory cytokine IL-8, even after Salmonella infection, and no difference was observed among the S.E-AvrA Ϫ , S.E-AvrA ϩ , or wild type S.E strains post-infection (Fig. 8E). Together, AvrA-induced suppression of the JNK pathway and TJ protein disassembly increases ZO-1 stability following S. Enteritidis infection.

Discussion
In the current study, we report that the S. Enteritidis effector protein AvrA stabilizes the TJ protein ZO-1, thus preserving intestinal epithelial permeability and decreasing bacterial invasion. Bacterial AvrA-mediated regulation of host TJs involves blocking the JNK signaling pathway, which was demonstrated in vitro and in vivo (Fig. 8F). The S. Enteritidis protein AvrA inhibits the JNK pathway in epithelial cells, enhances TJ barrier function by stabilizing ZO-1 expression at the protein level, increases cell permeability, reduces bacteria invasion, and reduces the inflammatory response in mice. The JNK inhibitor SP600125 abolished the stabilizing effect of AvrA on ZO-1.
Most Salmonella pathogenesis studies to date have focused on S. Typhimurium. In this study, we are the first to report that the S. Enteritidis effector protein AvrA inhibits the JNK pathway, thereby preventing intestinal epithelial barrier disruption, and normalizes the expression of the TJ protein ZO-1. There are numerous serotypes of Salmonella; different serotypes are found in different animal hosts or environments, and they may cause different symptoms in humans (6). Previous mouse studies have shown significant pathological changes in the intestine at 8 h after infection with wild type S. Typhimurium (20,21,37,44,45). In the current study, we found that mice did not show marked pathological or inflammatory changes 8 h after infection with S. Enteritidis. Instead, body weight loss occurred 1-2 days later than in S. Typhimurium colitis mice. We could detect significant increased serum IL-6 2 h after infection with wild type S. Typhimurium but not in the mice infection with S. Enteritidis for 8 h. All of the above results suggest that mice react differently when infected with S. Enteritidis or S. Typhimurium, despite nearly identical disease symptoms.
The JNK inhibitor SP600125 stabilizes epithelial barrier function by regulating claudin expression in murine and human epithelial cells (35,46). In the present study, we found that SP600125 enhanced TJ function by abolishing the effect of Salmonella AvrA on the TJ protein ZO-1. The Salmonella protein AvrA is an anti-inflammatory effector that possesses acetyltransferase activity toward specific host MAPKKs and inhibits the host JNK/AP-1 and NF-B signaling pathways (21)(22)(23)(24)(25). Previous studies on S. typhimurium showed that AvrA targets ZO-1 expression by blocking the JNK/AP-1 pathway and by targeting occludin expression (37,47). In the current study, we did not observe significant regulation of occludin by S. Enteritidis. Our previous study showed that AvrA of S. Typhimurium increased TJ proteins, including claudin-1 and claudin-2, but these claudins were not regulated by S. Enteritidis AvrA in this study. The adherens junction protein E-cadherin was decreased in the colon of mice treated with wild type S. Typhimurium (19). JNK also regulates adherens junctions and barrier function in keratinocytes (48). In the present study, E-cadherin levels were not changed in the mouse colon after S. Enteritidis infection. AvrA inhibited the JNK pathway and enhanced TJs by stabilizing ZO-1 but not the adherens junction protein E-cadherin. This indicates that S. Enteritidis AvrA enhances TJ barrier function by selectively regulating ZO-1 expression in intestinal epithelial cells, which is different from the role of S. typhimurium AvrA. There was no obvious change in ZO-1 expression at the transcriptional level after S. Enteritidis infection, but ZO-1 protein levels were noticeably changed. JNK activation leads to tyrosine phosphorylation of ZO-1 and the disruption of TJs. We speculate that AvrA-mediated stabilization of ZO-1 occurs at the post-translational level. Salmonella spp. possess a range of effector proteins; different serotypes may secrete different proteins, and the resulting interactions with effector proteins may differentially influence host cellular function. The observed differences between S. Enteritidis AvrA and S. Typhimu-

in Caco-2 BBE control cells and the disruption following infection by
AvrA-mutant bacteria. There were no significant changes in other TJ proteins, such as occludin, claudin-1, and claudin-7, or in the adhesion junction protein E-cadherin between S. E-AvrA Ϫ and the wild type or complemented strains. B, high TEER is evident in Caco-2 BBE cells after colonization with wild type or AvrA complemented strains. TEER was measured at the indicated times, and multiple comparisons of mean TEER values were performed using two-way ANOVA and generalized linear mixed models. The mean differences between the TEER values for the S.E-AvrA Ϫ mutant and either the S.E-AvrA ϩ or wild type strain were significant for follow-up measurements at 1-8 h (n ϭ 6). **, p Յ 0.01.
rium AvrA in terms of the regulation of TJ proteins may be due to interactions with different effectors.
Two p-JNK bands were detected by Western blotting. S. Enteritidis AvrA inhibited the 54-kDa p-JNK band to a greater extent than the 46-kDa p-JNK band, whereas S. Typhimurium AvrA inhibited both p-JNK bands. We also examined downstream molecules in the JNK pathway, including FoxOs and c-Jun. FoxO and total c-Jun levels were not changed by AvrA, whereas p-c-Jun was regulated by AvrA (data not shown). Similar to ZO-1, the activation of p-c-Jun by AvrA was abolished by treatment with the JNK inhibitor. Although Typhimurium AvrA targets multiple pathways, we found that the Enteritidis AvrA-mediated regulation of SPAK/JNK is specific.
Taken together, our data reveal a novel role of AvrA in S. Enteritidis in the stabilization of TJs and the attenuation of bacterial invasion in intestinal epithelial cells. We also identified differences between S. Enteritidis and S. Typhimurium, including delayed pathological changes during S. Enteritidis infection, different changes in tight junctions and adherens junction protein E-cadherin, and delayed responses in inflammatory cytokines. Bacterial effector proteins paralyze or reprogram eukaryotic cells to benefit pathogens. Our findings indicate an important role of S. Enteritidis effector AvrA, which influences host cellular function and bacterial invasion during infection and inflammation. The manipulation of JNK activity and TJs in microbial-host interactions may be a novel therapeutic approach for the treatment of infectious diseases.

Animals and Ethics Statement
C57BL/6 mice (female, 6 -8 weeks) were obtained from The Jackson Laboratory (Bar Harbor, ME). All animal work was approved by the Rush University and University of Illinois at Chicago Committee on Animal Resources. Euthanasia was accomplished via sodium pentobarbital administration (100 mg/kg body weight, i.p.) followed by cervical dislocation. All procedures were conducted in accordance with the approved guidelines of the Committees on Animal Resources.

Bacterial Strains and Growth Conditions
The S. Enteritidis WT strain C50336 was obtained from the Chinese National Institute for the Control of Pharmaceutical and Biological Products. The S. Enteritidis AvrA mutant strain S.E-AvrA Ϫ derived from C50336 and the AvrA complemented strain S.E-AvrA ϩ were constructed in this study. The S. typhi-murium wild type strain SL1344 (SB300) and the Salmonella AvrA mutant strain SB1117 derived from SL1344 were provided by Dr. Galan (24). The bacteria and plasmids used in this study are listed in Table 1. Bacterial growth conditions were as follows: non-agitated microaerophilic bacterial cultures were prepared by inoculating 10 ml of LB broth with 0.01 ml of a stationary phase culture, followed by overnight incubation (Ͼ18 h) at 37°C (49).

Construction of the AvrA Deletion Mutant and Complemented Strains
The AvrA deletion mutant was constructed using the -Redmediated recombination system, as described previously (50). The primers used in this study are listed in Table 2. Briefly, the chloramphenicol (Cm) cassette was amplified from the pKD3 plasmid using primers AvrA-cm forward and AvrA-cm reverse to include 39-bp homology extensions from the 5Ј and 3Ј ends of the AvrA gene. The PCR products were purified and introduced into the pKD46 plasmid containing S. Enteritidis C50336 by electroporation. Recombinant 50336-Cm bacteria were screened and selected on both Cm and ampicillin resistance LB agar plates. Allelic replacement of AvrA by the Cm cassette was verified by PCR screening (primers: AvrA-e forward and AvrA-e reverse) and DNA sequencing. Subsequently, the Cm cassette of 50336-Cm was excised by introducing the Flp recombinase-expressing vector pCP20. The full-length AvrA gene was cloned from S. Enteritidis strain C50336 using primers AvrA-r forward and AvrA-r reverse and ligated into plasmid pBR322. The recombined plasmid pBR322-AvrA was introduced into the AvrA mutant strain, and the transformants were screened and selected on ampicillin resistance LB agar plates. The AvrA mutant and complemented strains were confirmed by PCR, AvrA DNA sequencing, quantitative PCR, and Western blotting. Biochemical tests were performed using the API 20E identification kit (bioMerieux SA, Lyon, France) in accordance with the manufacturer's protocol.

Cell Treatment with the JNK Inhibitor SP600125
The JNK inhibitor SP600125 (50 mM; EMD Biosciences, San Diego, CA) was added directly to the culture medium 1 h before Salmonella treatment. Caco-2 BBE cells pretreated with SP600125 were incubated with Salmonella for 30 min, washed three times in Hanks' balanced salt solution (HBSS), incubated in DMEM containing 100 g/ml gentamicin and 50 M SP600125 for 30 min, and harvested. Protein levels were determined by Western blotting.

Streptomycin Pretreated Mouse Model
Water and food were withdrawn 4 h before oral gavage with 7.5 mg/mouse streptomycin. Afterward, animals were supplied with water and food. 20 h after streptomycin treatment, water   and food were once again withdrawn for 4 h before the mice were infected with 1 ϫ 10 8 CFU of Salmonella (100-l bacterial suspension in HBSS) or treated with sterile HBSS (control) by oral gavage, as previously described (42). At the indicated times post-infection, the mice were sacrificed, and intestinal tissue samples were removed for analysis.

Bacterial Colonization
Polarized human epithelial cells (Caco-2 BBE, HCT116, and SKCO15 cells) were grown in DMEM with 10% FBS. At 90 -100% confluence, the cells were colonized with an equal number of the indicated bacteria for 30 min, washed with HBSS, and incubated in DMEM containing gentamicin (100 g/ml) for 30 min. The first 30 min of incubation allowed bacteria to contact the epithelial cell surface and inject the effectors into the host cells. After extensive HBSS washing, extracellular bacteria were washed away. Incubation with gentamicin inhibited bacterial growth. To determine bacterial adhesion to cells, the cells were washed 30 min after colonization with prewarmed HBSS and then incubated with shaking in HBSS containing Triton X-100 (1%) in a cold room. To ascertain bacterial invasion, 30 min after bacterial colonization, the cells were incubated for an additional 30 min in DMEM with gentamicin, washed and incubated with shaking in HBSS containing Triton X-100 in a cold room. Bacterial CFU were determined by plating diluted cell lysates onto MacConkey agar culture plates (Difco Laboratories, Inc.) and incubating the cultures at 37°C overnight.

Immunoblotting for AvrA
Bacteria were lysed in lysis buffer (50v Tris, pH 8.0, 150v NaCl, 5 mM EDTA with a complete Mini protease inhibitor mixture (1 tablet/10 ml, Roche), and 1% Triton X-100), and sonicated. Equal amounts of total proteins were loaded, separated by SDS-PAGE, and processed for immunoblotting with custom-made AvrA antibody, as reported in our previous studies (19). The 15-amino acid peptide CGEEPFLPSD-KADRY was designed based on the AvrA sequence amino acids 216 -230 (GenBank TM accession no. AE008830).

Immunofluorescence
Colonic tissues from mice were freshly isolated and embedded in paraffin wax after fixation with 10% neutral buffered formalin. Immunofluorescence was performed on paraffinembedded sections (5 m). After preparation as previously described, the slides were incubated for 1 h in blocking solution (2% BSA and 1% goat serum in HBSS) to reduce nonspecific background. The tissue samples were incubated with the indicated primary antibodies, including ZO-1 (1:100, 33-9100), claudin-7 (1:100, 34-9100; Invitrogen), and E-cadherin (1:100, 610405; BD Biosciences), overnight at 4°C. Samples were then incubated with secondary antibodies (anti-mouse AF488, A-11029, anti-rabbit AF594, A-11037; Invitrogen), for 1 h at room temperature. Tissues were mounted with the SlowFade antifade kit (s2828; Life Technologies) and coverslipped, and the edges were sealed to prevent drying. Specimens were examined with a Zeiss laser scanning microscope 710 (Carl Zeiss, Inc., Oberkochen, Germany). Fluorescence images were analyzed for ZO-1 distribution using image analysis software (LSM 710 META, version 4.2; Carl Zeiss). Each analysis was performed in triplicate for each tissue section on a total of 10 images/mouse sample (n ϭ 5).

Quantitative Pathological Analysis
Segments of the cecum and colon were fixed and embedded in paraffin according to standard procedures. Cryosections (5 m) were mounted on glass slides, air dried for 2 h at room temperature, and stained with hematoxylin and eosin (H&E). Pathological evaluations were performed by two pathologists in a blinded manner. Based on an earlier study (42,60), a scoring scheme for quantitative pathological analysis of cecal inflammation in H&E-stained sections (5 m) was followed.
Submucosal Edema-Submucosal edema was scored as follows: 0, no pathological changes; 1, mild edema (submucosa: 0.20-mm-wide and accounting for 50% of the diameter of the entire intestinal wall (tunica muscularis to epithelium)); 2, moderate edema (submucosa: 0.21-0.45-mm-wide and accounting for 50 -80% of the diameter of the entire intestinal wall); and 3, profound edema (submucosa: 0.46-mm-wide and accounting for 80% of the diameter of the entire intestinal wall). Submucosa widths were determined by quantitative microscopy and represent the average of 30 evenly spaced radial measurements of the distance between the tunica muscularis and the lamina muscularis mucosae.
PMN Infiltration into the Lamina Propria-Polymorphonuclear granulocytes (PMNs) in the lamina propria were enumerated in 10 high power fields (400ϫ magnification; field diame-ter, 420 m), and the average number of PMNs/high power field was calculated. The scores were defined as follows: 0, Ͻ5 PMNs/high power field; 1, 5-20 PMNs/high power field; 2, 21-60 PMNs/high power field; 3, 61-100 PMNs/high power field; and 4, Ͼ100 PMNs/high power field. Transmigration of PMNs into the intestinal lumen was consistently observed for situations with Ͼ60 PMNs/high power field.
Goblet Cells-The average number of goblet cells per high power field (magnification, 400ϫ) was determined in 10 different regions of the cecal epithelium. Scoring was as follows: 0, Ͼ28 goblet cells/high power field (magnification, 400ϫ; in the cecum of normal SPF mice, we observed an average of 6.4 crypts/high power field, and the average crypt consisted of 35-42 epithelial cells, of which 25-35% were differentiated into goblet cells); 1, 11-28 goblet cells/high power field; 2, 1-10 goblet cells/high power field; and 3, Ͻ1 goblet cell/high power field.
Epithelial Integrity-Epithelial integrity was scored as follows: 0, no detectable pathological changes in 10 high power fields (400ϫ magnification); 1, epithelial desquamation; 2, erosion of the epithelial surface (gaps of 1-10 epithelial cells/lesion); and 3, epithelial ulceration (gaps of 10 epithelial cells/ lesion; at this stage, there is generally granulation tissue below the epithelium). We averaged the two independent scores for submucosal edema, PMN infiltration, goblet cells, and epithelial integrity for each tissue sample. The combined pathological score for each tissue sample was determined as the sum of these averaged scores. The combined pathological score ranged between 0 and 13 arbitrary units and covered the following levels of inflammation: 0, intestinal intact without any signs of inflammation; 1-2, minimal signs of inflammation (this was frequently found in the ceca of SPF mice; this level of inflammation is generally not considered a sign of disease); 3-4, slight inflammation; 5-8, moderate inflammation; and 9 -13, profound inflammation.

IL-8 Enzyme-linked Immunosorbent Assay
IL-8 was measured in cell culture medium using a sandwich ELISA method. Briefly, goat anti-human IL-8 antibody (1 g, AB-208-NA; R&D Systems) was coated on plastic wells at 4°C overnight. Then culture medium and recombination human IL-8 standards were added to each well. After washing, the plates were incubated with rabbit anti-human IL-8 antibody (1 g, ab7747; Abcam, Cambridge, UK), washed again, and then incubated with peroxidase-conjugated anti-rabbit IgG antibody (214-1516; KPL Inc., Gaithersburg, MD). Following complete washing, KPL SureBlue peroxidase substrate (52-00-01) was added. The A 450 values were determined after stop solution was added to the wells.

Real Time Quantitative PCR Analysis
Total RNA was extracted from mouse epithelial cells or cultured cells using TRIzol reagent (Invitrogen). RNA integrity was verified by electrophoresis. Reverse transcription of RNA was performed using the iScript cDNA synthesis kit (Bio-Rad) according to the manufacturer's protocol. The cDNA was subjected to quantitative real time PCR using a CFX Connect real time system (Bio-Rad) and SYBR Green Supermix (Bio-Rad).
All expression levels were normalized to ␤-actin levels in the same sample. The percentage of expression was calculated as the ratio of the normalized value of each sample to that of the corresponding untreated control sample. All real time PCR analyses were performed in triplicate.

Statistical Analysis
All data are presented as the means Ϯ S.D. All statistical tests were two-sided. p values of less than or equal 0.05 were considered statistically significant. Differences between two samples were analyzed using Student's t test. Multiple comparisons of mean TEER (Fig. 3B) were performed using two-way ANOVA, and p values were adjusted using Tukey's method. Because we were interested in comparing the mean differences among the wild type, S.E-AvrA ϩ and S.E-AvrA Ϫ groups, the group effects at each time point of 0, 0.5, 1, 2, 3, 4, 5, 6, 7, or 8 h were tested using generalized linear mixed models. The pairwise treatment group differences among the wild type, S.E-AvrA ϩ and S.E-AvrA Ϫ groups at each time point were compared using Tukey's method to adjust p values to ensure accurate results. For fluorescence intensity data, one-way ANOVA was conducted to analyze the mean differences among the control, S.E-AvrA ϩ , wild type, and S.E-AvrA Ϫ groups, and the p values were adjusted using Tukey's method. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC).
Author Contributions-Z. L. and Y.-G. Z. were responsible for data acquisition, analysis, and interpretation and drafting of the manuscript. Z. L. and J. S. wrote the main manuscript text and prepared figures. Y. X. was responsible for statistical analysis and drafting of the manuscript. X. J. and X. X. were responsible for technical or material support and drafting of the manuscript. J. S. was responsible for the study concept and design, critical revision of the manuscript for important intellectual content, and study supervision. All authors reviewed the manuscript.