ABSTRACT
Enterotoxigenic Escherichia coli (ETEC) is the leading cause of traveler's diarrhea and children's diarrhea worldwide. Among its virulence factors, ETEC produces heat-labile enterotoxin (LT). Most secreted LT is associated with outer membrane vesicles that are rich in lipopolysaccharide. The majority of prior studies have focused on soluble LT purified from ETEC periplasm. We investigated the hypothesis that the extracellular vesicle context of toxin presentation might be important in eliciting immune responses. We compared the polarized epithelial cell responses to apically applied soluble LT and LT-containing vesicles (LT+ vesicles) as well as controls using a catalytically inactive mutant of LT and vesicles lacking LT. Although vesicle treatments with no or catalytically inactive LT induced a modest amount of interleukin-6 (IL-6), samples containing catalytically active LT elicited higher levels. A combination of soluble LT and LT-deficient vesicles induced significantly higher IL-6 levels than either LT or LT+ vesicles alone. The responses to LT+ vesicles were found to be independent of the canonical LT pathway, because the inhibition of cyclic AMP response element (CRE)-binding protein (CREB) phosphorylation did not lead to a decrease in cytokine gene expression levels. Furthermore, soluble LT caused earlier phosphorylation of CREB and activation of CRE compared with LT+ vesicles. Soluble LT also led to the activation of activator protein 1, whereas LT+ vesicle IL-6 responses appeared to be mediated by NF-κB. In summary, the results demonstrate that soluble LT and vesicle-bound LT elicit ultimately similar cytokine responses through distinct different activation pathways.
INTRODUCTION
Enterotoxigenic Escherichia coli (ETEC) is the leading cause of traveler's diarrhea (3), and it has been estimated to cause approximately 10 million cases of traveler's diarrhea worldwide (45). ETEC is also the leading cause of morbidity and mortality due to diarrhea in children in developing countries. A total of 280 million cases of diarrhea associated with ETEC were found in children less than 5 years old in outpatient clinics in developing countries (50), and mortality due to ETEC has been estimated at 170,000 deaths annually (32). Heat-labile enterotoxin (LT) is a major virulence factor produced by ETEC and is known to contribute to the disease (20).
LT is an AB5 toxin that is composed of a pentameric B subunit, which binds to host receptors, and a catalytically active A subunit (12). The B pentameric ring binds to the Galβ1,3GalNAcβ1(NeuAcα2,3),4Galβ1,4Glc ceramide (GM1) ganglioside on host cells, which mediates internalization. Studies using soluble LT that was purified from the periplasm have led to a detailed understanding of its complex trafficking pathway and activation inside mammalian cells (12, 20, 39). Once internalized, LT is trafficked to the Golgi apparatus and the endoplasmic reticulum (ER), where the A subunit is further processed. The modified A subunit then catalyzes ADP-ribosylation of the Gsα subunit in the adenylate cyclase pathway. This ribosylation leads to an increase in cyclic AMP (cAMP) levels and an efflux of water and electrolytes into the lumen of the intestine (12). LT activates protein kinase A (PKA), which phosphorylates the cystic fibrosis transmembrane regulator (CFTR) and cAMP response element (CRE)-binding protein (CREB) to transport Cl− into the intestinal lumen and induce gene transcription (40).
In contrast to the highly homologous cholera toxin (CT) produced by Vibrio cholerae, most of the secreted LT is found associated with outer membrane vesicles (OMVs) (14, 18, 44). OMVs are spherical structures secreted from all Gram-negative bacteria studied to date (26). OMVs are enriched in outer membrane components, and the lumen of OMVs contains periplasmic components. OMVs contain biologically active components and immunomodulatory molecules (also known as pathogen-associated molecular patterns [PAMPs]), such as lipopolysaccharide (LPS) and flagellin, that interact with and influence host cells (10). LT is found in the OMV lumen and, by virtue of its ability to bind LPS, also bound to the surface of OMVs (17, 31).
Besides LT, other toxins from a variety of pathogenic bacteria have also been shown to be enriched in OMVs, including cytolysin A (ClyA) from E. coli (49), leukotoxin A (LktA) from Aggregatibacter actinomycetemcomitans (21), vacuolating toxin (VacA) from Helicobacter pylori (11), and cytolethal distending toxin (CDT) from Campylobacter jejuni (29). Previous studies have demonstrated differences in mammalian cell toxicity based on the presentation of those toxins (e.g., in the context of OMVs versus soluble toxin). Wai et al. showed that compared to equal amounts of ClyA purified from the periplasm, OMV-associated ClyA induced higher cytotoxic activity in HeLa cells (49). The OMV context of ClyA presentation was shown to facilitate the active oligomerized state of ClyA, leading to its higher activity (49). H. pylori has been shown to secrete VacA both in a free soluble form and associated with OMVs (11). Ricci et al. demonstrated that although OMV-associated VacA accounted for approximately 25% of secreted toxin, the OMV-bound toxin showed lower vacuolating activity than soluble VacA (36). However, as OMVs are complex entities, the OMV association of toxins is likely to affect host cells beyond the differences in toxin potency.
The gut is a unique environment that has developed tolerance to native microbiota. The intestinal epithelium forms tight intracellular junctions and microvilli that inhibit the attachment and invasion of intestinal organisms (1). Tight junctions also play a role in the polarization of epithelial cells, resulting in distinct apical and basolateral surfaces (24). Commensal tolerance is further maintained through a variety of mechanisms, including the subcellular localization of Toll-like receptors (TLRs) and the inhibition of immune responses to commensal products, such as LPS (1). Pathogenic bacteria subvert these defenses in a variety of ways, including epithelial cell internalization and the elaboration of virulence factors, such as toxins (37). OMVs can also penetrate the epithelial cell barrier. ETEC OMVs were found to enter cultured intestinal epithelial cells via a specific LT-mediated pathway (23).
We hypothesized that, in addition to modulating its toxicity, the context of toxin presentation is important in determining the host response. In particular, OMV-associated toxin is likely to elicit a different inflammatory response than soluble toxin because of the presence of LPS and other PAMPs. To investigate this theory, we compared the responses of polarized human intestinal epithelial cells to apically applied soluble LT, catalytically inactive LT (S63K), OMVs containing either catalytically active LT (LT+ OMVs) or catalytically inactive LT (S63K OMVs), and OMVs without LT (ΔLT OMVs). We found significant differences in the kinetics of responses induced by soluble LT and by OMV-bound LT. Our results show that soluble LT and LT+ OMVs elicit different responses and act through different mechanisms.
MATERIALS AND METHODS
Strains and growth conditions.The bacterial strains used in this study are listed in Table 1. Strains were grown in CFA medium (1% Casamino Acids, 0.15% yeast extract, 0.005% MgSO4 and 0.005% MnCl2) at 37°C with or without 100 μg/ml ampicillin. The LT S63K mutants were constructed using the QuikChange mutagenesis kit (Qiagen) according to the manufacturer's instructions and using the following primers: S63K sense, 5′-GACGGATATGTTTCCACTAAACTTAGTTTGAGAAGTGC-3′, and S63K antisense, 5′-GCACTTCTCAAACTAAGTTTAGTGGAAACATATCCGTC-3′. Transformations were performed using a CaCl2 protocol, as previously described (17). To induce the expression of plasmid-encoded wild-type and mutant LTs, 100 μM isopropyl β-d-1-thiogalactopyranoside was added to cultures.
Strains used in this study
Cell culture.The human intestinal epithelial T84 cell line (American Type Culture Collection CCL-248) was maintained in a 1:1 ratio of Dulbecco's modified Eagle's medium and Ham's F12 medium (Gibco) supplemented with 10% fetal bovine serum (FBS) (HyClone) and 1% penicillin-streptomycin-amphotericin B (Gibco) at 37°C. Human embryonic kidney 293T (HEK293T) cells (ATCC CRL-11268) were maintained in minimum essential medium supplemented with 10% FBS at 37°C. For polarization assays, 4 × 105 T84 cells were seeded on 1.12-cm2 transwell inserts (Corning) and grown for 5 to 7 days. Tight-junction formation was measured using transepithelial electrical resistance (TEER) with a Millicell-ERS (Millipore), and cells were used at TEERs of ≥1,000 Ω·cm2. Cells were used between passages 65 and 73.
LT purification.LT was purified from strains containing wild-type LT or S63K LT as described previously (31). The concentration of LT was quantified using the Bradford method, with bovine serum albumin as the standard. The endotoxin concentration of LT was determined using a Limulus amebocyte lysate assay (Cambrex) and was found to be less than 1 endotoxin unit (EU)/ml for both wild-type LT and S63K LT in the concentrations used in the assays.
OMV purification and standardization.OMVs were purified from strains expressing wild-type LT or S63K LT or from the ΔeltA strain as previously described (22). To quantify the concentration of LT in LT+ OMVs, 2-fold dilutions of LT were loaded on a 15% SDS-polyacrylamide gel with dilutions of LT+ OMVs. Proteins were transferred to a polyvinylidene fluoride membrane (GE Healthcare), incubated with an LT-cross-reactive rabbit polyclonal anti-CT antibody (Sigma), incubated with a horseradish peroxidase-conjugated anti-rabbit antibody (Sigma), and visualized using enhanced chemiluminescence (SuperSignal; Pierce). A standard curve was calculated from the known concentrations of the soluble LT and used to calculate the amount of LT in OMVs using densitometry values obtained with ImageJ software (National Institutes of Health). S63K and ΔLT OMVs were standardized to LT+ OMVs by lipid content using standard curves generated using the lipophilic dye FM4-64 (Molecular Probes). Consequently, for example, in an experiment using “1 nM samples,” treatments were 1 nM soluble LT (WT or S63K), OMV preparations containing 1 nM LT (LT+ OMVs and LT-supplemented ΔLT OMVs), and the equivalent amount of OMVs that did not contain wild-type LT (S63K OMVs and ΔLT OMVs).
Signal pathway inhibitors.Inhibitors of PKA (adenosine 3′,5′-cyclic phosphorothioate-Rp; Rp-camps), p38 [4-(4-fluorophenyl)-2-(4-methylsulfinylphenyl)-5-(4-pyridyl)1H-imidazole; SB203580], Jun N-terminal protein kinase (JNK) {anthra[1,9-cd]pyrazol-6(2H)-one [1,9-pyrazoloanthrone]; SP600125}, MEK (2′-amino-3′-methoxyflavone; PD98059), and NF-κB (ammonium pyrrolidinedithiocarbamate [PDTC]) were purchased from Calbiochem. The CREB inhibitor H89 {N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide dihydrochloride} was purchased from Sigma-Aldrich. SB203580 (10 μM), SP600125 (10 μM), PD98059 (10 μM), and H89 (10 μM) were added bilaterally to polarized cells for 1 h, Rp-camps (100 μM) was added basolaterally for 1 h, and PDTC (10 μM) was added bilaterally for 30 min before the addition of treatments.
RT-PCR.Samples standardized to either 1 nM or 200 pM LT or vehicle (phosphate-buffered saline [PBS]) were added to the apical compartment of duplicate wells of polarized T84 cells for 6 h with or without inhibitor pretreatment. Total RNA was collected using the Qiagen RNeasy kit according to the manufacturer's instructions. The RNA was reverse transcribed into cDNA using oligo(dT) primers and SuperScript III (Invitrogen), according to the manufacturer's instructions, and the cDNA was used as the template in reverse transcription-PCR (RT-PCR) assays. RT-PCR was performed in a total volume of 15 μl using iQ SYBR green (Bio-Rad) and analyzed using an iCycler real-time detection system (Bio-Rad). The gene-specific primers used for RT-PCR analysis were as follows: interleukin-6 (IL-6) sense, 5′-GACAGCCACTCACCTCTT-3′; IL-6 antisense, 5′-TGTTTTCTGCCAGTGCC-3′; tumor necrosis factor alpha (TNF-α) sense, 5′-CCCAGGCAGTCAGATCAT-3′; TNF-α antisense, 5′- TCAGCTCCACGCCATT-3′; glyceraldehyde-3-phosphate dehydrogenase (GAPDH) sense, 5′-ACATCGCTCAGACACCAT-3′; and GAPDH antisense, 5-′GGGTCATTGATGGCAACA-3′. Results were analyzed using the accompanying software, and gene expression was standardized to GAPDH levels. Results are shown as the fold induction of the gene in relation to the corresponding mock treatments and were measured using the 2−ΔΔCT method.
ELISA.Polarized T84 cells were treated apically with samples for 10 h, and both apical and basal supernatants were collected and analyzed using IL-6 and TNF-α enzyme-linked immunosorbent assay (ELISA) kits (BD Biosciences), according to the manufacturer's instructions.
Flagellin immunoblotting.The MV samples were analyzed using 10% SDS-PAGE (Bio-Rad) and transferred to a nylon membrane. The membrane was then blocked in Odyssey blocking buffer and incubated with rabbit anti-H7 (a flagellar marker originally produced by Difco, a kind gift from Patrick Seed, Duke University). The membrane was then incubated with a fluorescently conjugated anti-mouse antibody and imaged using the Odyssey imaging system.
Immunoblotting of nuclear extracts.Samples were added to polarized T84 cells in the presence or absence of inhibitors for the indicated times, and nuclear protein was extracted as previously described (42), with some modifications. Briefly, buffer A (10 mM HEPES [pH 8], 1.5 mM MgCl2, 10 mM KCl, and 0.5 mM dithiothreitol [DTT]) supplemented with 0.4% Igepal was directly added to cells in the transwell insert and incubated on ice for 15 min. Cells were then collected, vortexed for 15 s, and centrifuged at 8,000 × g for 2 min. The supernatant was discarded, and the nuclear pellet was washed with buffer A. The nuclear pellet was then resuspended in buffer B (20 mM HEPES [pH 7.9], 1.5 mM MgCl2, 0.42 M NaCl, 0.2 mM EDTA, 25% glycerol, 0.5 mM DTT, and protease and phosphatase inhibitor cocktails [Sigma-Aldrich]) and incubated at 4°C for 2 h with shaking. The supernatant was then desalted using a desalting column (Thermo Scientific) and analyzed using 10% SDS-PAGE. The proteins were transferred to a nitrocellulose membrane and incubated with Odyssey blocking buffer before being incubated with mouse anti-TATA-binding protein (TBP) (Abcam), mouse anti-pCREB (Millipore), rabbit anti-pMSK1 (S376) and rabbit anti-pMSK1 (T581) (Cell Signaling), and rabbit anti-pp65 S276 (Abcam). To ensure that the nuclear preparations were free of cytosolic contamination, the membranes were also incubated with mouse anti-β-tubulin. The β-tubulin monoclonal antibody developed by Michael Klymkowsky was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA. The membranes were then incubated with the appropriate fluorescently conjugated secondary antibodies and visualized using the LiCor Odyssey imaging system. No β-tubulin bands were detected in the nuclear extracts, indicating that they were free from cytosolic contamination. The densitometry of the TBP and pCREB bands was determined using the accompanying Odyssey software. Results are presented as the pCREB/TBP ratio normalized to a mock value of 1 for each blot.
Luciferase assay.A total of 1 × 104 HEK293T cells were seeded into each well of a 96-well plate 16 h before transfection. Cells were then transfected with the pCRE-Luc plasmid (Stratagene) containing CRE fused to a firefly luciferase reporter gene and the pSV40-RL plasmid (Promega) containing a Renilla luciferase reporter fused to a constitutive promoter using Lipofectamine 2000 (Invitrogen). Samples were added at approximately 16 h posttransfection for the indicated time, and firefly luciferase and Renilla luciferase activities were determined using the Dual-Glo luciferase assay (Promega), according to the manufacturer's instructions. The results are presented as the ratio of firefly luciferase activity to Renilla luciferase activity.
EMSA.Nuclear extracts were prepared as described previously. Extracts were then mixed with an equal volume of buffer C (20 mM HEPES [pH 7.9], 50 mM KCl, 0.2 mM EDTA, 0.5 mM DTT, protease inhibitor cocktail and phosphatase inhibitor cocktail), and the protein concentration was measured using the Bradford method. Nuclear protein was analyzed using an electrophoretic mobility shift assay (EMSA) to determine the activation of activator protein 1 (AP-1) according to the manufacturer's instructions (Gel Shift Assay System; Promega). Phosphorimages were scanned using the STORM 860 system (Molecular Dynamics) and analyzed using ImageQuant 5.2 software (Molecular Dynamics).
Statistical analysis.Multiple comparisons were performed using one-way analysis of variance (ANOVA) followed by Tukey's test. Comparisons between treated and untreated samples were performed using Student's t test. All statistical analyses were performed using GraphPad InStat (GraphPad Software, Inc.). Results are presented as means ± standard errors of the means (SEM), and all experiments were performed in triplicate.
RESULTS
Treatments containing catalytically active LT induce higher expressions of IL-6 and TNF-α than other treatments.To investigate whether the context of LT affects the toxicity and host response to LT, we treated polarized cultures of human T84 intestinal epithelial cells with either soluble LT or equivalent concentrations of LT in the context of OMVs. Cytokine induction in response to LT and OMVs, individually, has been widely studied using a variety of cultured mammalian cells (4, 5, 10, 41, 43). However, no studies on the pathways involved in cytokine responses have been performed using the more physiologically relevant polarized epithelial cell model. As a result, these experiments will also reveal host cell responses to strictly apical application of LT.
We designed a series of samples that would allow us to distinguish the effects of the OMVs, the effects of active toxin, and the effects of the combination of OMVs with active toxin. The samples were carefully standardized to allow comparative evaluations to be made. The amount of LT in LT+ OMVs was first standardized to the amount of soluble LT. S63K OMVs and ΔLT OMVs were then standardized to LT+ OMVs by their lipid content. Comparing equivalent treatments of LT+ OMVs and ΔLT OMVs was more appropriate than using pure LPS as an endotoxin control because the presentation of LPS in OMVs is very different from that in preparations of organically extracted, sonicated LPS. In addition, we combined soluble LT and ΔLT OMVs (LT/ΔLT OMVs) in proportions equivalent to their concentrations in LT+ OMV preparations to determine if the response induced by LT+ OMVs was merely additive or whether the native context of toxin presentation played a role.
As shown in Fig. 1 A, whereas all treatments induced TNF-α expression compared with mock, soluble LT, LT+ OMVs and the combination of LT/ΔLT OMVs induced higher TNF-α gene expression levels than the other treatments. This induction appeared to be dependent on the catalytic activity of LT, because the catalytically inactive S63K LT treatments and the ΔLT OMVs did not induce TNF-α to the same level, even at the higher 1 nM dose (Fig. 1A). We noted that the amount of TNF-α induction was not significantly different between soluble and vesicle-associated LT.
Treatments containing catalytically active LT elicit high levels of TNF-α and IL-6 expression in polarized intestinal epithelial cells, and induction is independent of flagellin. (A and B) Polarized T84 cells were treated with 200 pM or 1 nM standardized amounts (see Materials and Methods) of LT, LT+ OMVs, ΔLT OMVs, LT/ΔLT OMVs, S63K LT, or S63K OMVs as indicated, and the TNF-α (A) and IL-6 (B) gene expression levels were measured at 6 h using RT-PCR. (C) Polarized T84 cells were incubated with samples standardized to 1 nM LT for 10 h, and IL-6 levels were measured in supernatants collected from the apical compartment using ELISA. (D) Flagellin was detected in preparations of LT+ OMVs, ΔLT OMVs, LT/ΔLT OMVs, and S63K OMVs by immunoblotting with anti-FliC. The preparations were concentrated to allow for detection of flagellin, but the concentrations remained proportional to those used in the treatments described above.
As shown in Fig. 1B, the treatments containing catalytically active LT induced higher IL-6 gene expression levels than the other treatments at both 1 nM and 200 pM. In addition, the two concentrations of LT induced equivalent amounts of IL-6, suggesting that the 200 pM treatment already produced the maximal effect. Both ΔLT and S63K OMVs induced a modest amount of IL-6, whereas S63K LT did not induce expression over mock. In contrast to the case for TNF-α, whose expression was not significantly different with the different LT-containing treatments, the combination of LT and ΔLT OMVs led to significantly higher IL-6 levels than both soluble LT and LT+ OMVs at both concentrations (by ANOVA, P < 0.01 at 1 nM and P < 0.001 at 200 pM). This induction was synergistic, as the level of induction was more than the combined results with LT and ΔLT OMVs. Notably, again, the amount of IL-6 gene induction was not significantly different between soluble LT and LT+ OMVs at either concentration.
The IL-6 ELISA results in Fig. 1C confirmed that the induction of gene expression also resulted in an increase in protein concentration in the apical compartment. No IL-6 was found in the basolateral supernatants, indicating the polarized secretion of this cytokine (data not shown). TNF-α protein expression was not found in either compartment of treated cell cultures, which may be due to an extremely low, undetectable basal level of TNF-α expression by these cells. Overall, these results indicated that, independent of its context as a soluble molecule or as a component of OMVs, equivalent amounts of catalytically active LT elicited substantial but indistinguishable TNF-α and IL-6 upregulation and IL-6 secretion in polarized epithelial cells.
We tested whether levels of flagellin in the OMV preparations directly correlated with the observed induction levels, since flagellin is known to induce cytokine expression and to be associated with OMVs (13, 33, 46). We analyzed the relative amounts of flagellin associated with the OMV treatment preparations using immunoblotting. As shown in Fig. 1D, LT+ OMVs and S63K OMVs had similar levels of flagellin, and these were substantially higher than the amount of flagellin present in ΔLT OMVs. Thus, although the flagellin present in OMVs might contribute to the induction of TNF-α and IL-6, the levels of flagellin did not correlate with the differences in the levels of induction observed for the different OMV samples.
LT+ OMVs do not induce cytokine gene expression through CREB.To further examine the role of the catalytic activity of LT in inducing IL-6 and TNF-α gene expression, we used Rp-camps and H89 to examine whether the induction occurs through PKA. As mentioned above, LT acts by increasing cAMP levels, which activates PKA, leading to the phosphorylation of CREB. Rp-camps prevents the dissociation of the catalytic subunits of PKA from their regulatory subunits, abolishing PKA activity, and H89 inhibits the phosphorylation of CREB by PKA (30). As expected, Rp-camps significantly inhibited the induction of TNF-α and IL-6 in response to soluble LT, as determined at 6 h posttreatment (Fig. 2 A and B). However, Rp-camps inhibition of PKA activity notably did not have an effect on the responses induced by LT+ OMVs at 6 h (Fig. 2A and B). The combination of LT/ΔLT OMVs showed an intermediate response to inhibition by Rp-camps, since Rp-camps pretreatment significantly inhibited IL-6 gene induction but had no effect on TNF-α gene induction. Pretreatment with the H89 inhibitor led to a significant and more substantial reduction of IL-6 expression in response to soluble LT and the combination of LT/ΔLT OMVs but showed no effect on IL-6 gene induction in response to LT+ OMVs (Fig. 2B). However, whereas H89 showed effects similar to those of Rp-camps in TNF-α induction for most treatments, it led to a significant inhibition of TNF-α expression in response to LT+ OMVs (Fig. 2A). In sum, it was notable that differences in the presentation of similar concentrations of LT (LT+ OMVs compared with soluble LT) resulted in differences in the activation of PKA.
TNF-α and IL-6 induction in response to LT+ OMVs is independent of CREB phosphorylation. (A and B) Polarized T84 cells were pretreated with 100 μM Rp-camps or 10 μM H89 for 1 h and incubated with samples standardized to 200 pM LT. After 6 h, the levels of TNF-α (A) and IL-6 (B) gene induction were evaluated using RT-PCR. Asterisks indicate significant differences from the respective untreated sample: *, P < 0.05; **, P < 0.01; ***, P < 0.005. (C) Nuclear extracts were prepared from polarized T84 cells incubated with samples standardized to 1 nM LT for 4 h (upper panel) or 6 h (lower panel), and immunoblotting was performed to determine the phosphorylation status of CREB. TBP was used as a loading control. The image shown is representative of three independent experiments. (D) Densitometry analysis of three independent immunoblots for each time point. Asterisks indicate significant differences from the respective mock control at the corresponding time: *, P < 0.05; **, P < 0.01; ***, P < 0.005. (E) HEK293T cells were cotransfected with a firefly luciferase reporter fused to a CRE promoter and a Renilla luciferase reporter fused to a constitutively active promoter. Transfected cells were then incubated with samples standardized to 1 nM LT for 4 h and 6 h, and the firefly luciferase activity was measured and normalized to Renilla luciferase values. Asterisks indicate significant differences from the respective mock control at the corresponding time: *, P < 0.05; **, P < 0.01; ***, P < 0.005.
To examine the activation of CREB, immunoblotting of nuclear extracts was performed to determine the level of pCREB present in cells incubated with our samples, and the blots were quantitatively analyzed by densitometry. The immunoblots showed that at 4 h, CREB was phosphorylated in response to soluble LT and the combination of LT/ΔLT OMVs but was not phosphorylated in response to LT+ OMVs (Fig. 2C, upper panels). However, when assayed 2 h later, CREB was phosphorylated in response to LT+ OMVs as well as soluble LT and the combination of LT/ΔLT OMVs (Fig. 2C, lower panels). Densitometry analysis (Fig. 2D) showed that pCREB was significantly increased (approximately 2-fold) in response to LT and LT/ΔLT OMVs compared to the other treatments at 4 h and increased to approximately 3.75-fold at 6 h. LT+ OMV treatment led to substantially higher CREB phosphorylation than mock at 6 h, although this increase was not statistically significant. At both time points, both soluble LT and the combination of LT/ΔLT OMVs induced significantly higher levels of pCREB than LT+ OMVs. Therefore, treatment with LT in its native OMV-associated state appeared to activate CREB more slowly than treatments containing soluble LT.
We next used an independent assay to further examine the differences in the kinetics of cAMP-dependent activation. We compared the effects of soluble LT and OMV-bound LT on the activation of CRE genetic elements using a dual-luciferase reporter system. The CRE element is activated in response to cAMP induction, and therefore this assay provides a more sensitive detection of the downstream effects of cAMP production. Because we found it technically unfeasible to transfect T84 cells and maintain their ability to form a polarized monolayer, we used HEK293T cells for these assays. HEK293T cells were transfected with a reporter plasmid containing the CRE promoter fused to firefly luciferase as a reporter. To control for transfection efficiency, cells were cotransfected with a plasmid that constitutively expressed Renilla luciferase.
The CRE reporter results were consistent with the kinetics of CREB activation (Fig. 2E). Although LT+ OMVs stimulated CRE activity approximately 5-fold at 4 h, this induction was not significantly above that in mock-treated cells (P > 0.05). In comparison, after only 4 h, both soluble LT and the combination of LT/ΔLT OMVs significantly induced CRE activity above mock (approximately 15- and 22-fold higher than mock, respectively), and the response to the combination of LT/ΔLT OMVs was significantly higher than that for the same amount of soluble LT (P < 0.01). At 6 h, CRE activity was significantly induced by all of the treatments containing active LT: soluble LT, LT+ OMVs, and the combination of LT/ΔLT OMVs (approximately 35-, 19-, and 28-fold, respectively). These results confirm that the LT present in native LT+ OMVs causes the phosphorylation of CREB and activation of CRE but that this is delayed. However, as PKA pathway inhibition did not decrease LT+ OMV-mediated induction of IL-6 and TNF-α gene expression (Fig. 2A and B), it should be noted that the activation of CREB cannot account for the increased expression levels of these genes that are induced by LT+ OMVs.
The induction of TNF-α, but not that of IL-6, depends on ERK1/2 and p38 MAP kinases.We next investigated pathways other than CREB activation through which LT+ OMVs could induce the expression of TNF-α and IL-6. Mitogen-activated protein (MAP) kinases, such as p38, ERK1/2, and JNK, have long been known to play a role in the induction of cytokines in response to stimuli (2, 15, 19). Specific inhibitors of MEK1/2 and p38 were used to determine whether ERK1/2 and p38, respectively, played a role in the induction of TNF-α and IL-6 by LT and OMVs.
We found that p38 and ERK1/2 had similar roles in the induction of TNF-α and that neither played a role in IL-6 induction. MEK inhibition significantly reduced the induction of TNF-α by soluble LT and LT+ OMVs and substantially reduced its induction by the combination of LT/ΔLT OMVs (Fig. 3 A). p38 inhibition also significantly reduced the induction of TNF-α by soluble LT and LT+ OMVs but, notably, did not affect the response to the combination of LT/ΔLT OMVs (Fig. 3B). Neither inhibitor significantly reduced the induction of IL-6 by soluble LT, LT+ OMVs, or ΔLT OMVs (Fig. 3C and D). Interestingly, p38 and ERK1/2 seemed to play, if anything, opposite roles in the induction of IL-6 in response to the combination of LT/ΔLT OMVs. Inhibition of ERK1/2 resulted in a significant decrease in IL-6, whereas p38 inhibition significantly upregulated IL-6 expression (Fig. 3C and D). Together, these data demonstrate that both ERK1/2 and p38 play a role in the induction of TNF-α in response to both LT and LT+ OMVs, but the induction of IL-6 in response to these treatments is independent of these MAP kinases.
MEK1/2 and p38 play similar roles in the induction of TNF-α in response to both soluble LT and LT+ OMVs but do not play a role in IL-6 induction. Polarized T84 cells were pretreated for 1 h with inhibitors of MEK1/2 (10 μM PD98059) or p38 (10 μM SB203580) before being incubated with samples standardized to 200 pM LT for 6 h. (A and B) TNF-α levels were measured using RT-PCR after pretreatment with inhibitors of MEK1/2 (A) and p38 (B). (C and D) IL-6 levels were measured using RT-PCR after pretreatment with inhibitors of MEK1/2 (C) and p38 (D). Asterisks indicate significant differences from the corresponding untreated sample: *, P < 0.05; **, P < 0.01.
AP-1 is involved in the induction of TNF-α and IL-6 in response to soluble LT but not LT+ OMVs.We also determined the role of the JNK MAP kinase pathway in the induction of TNF-α and IL-6 in response to soluble LT and LT+ OMVs. JNK pathway inhibition led to a significant decrease in the level of expression of TNF-α in response to soluble LT, but it did not significantly inhibit TNF-α induction for any of the other treatments (Fig. 4 A). JNK pathway inhibition also led to a significant decrease in the level of expression of IL-6 in response to soluble LT (Fig. 4B). However, in contrast to the case for TNF-α, IL-6 expression levels induced in response to ΔLT OMVs and the combination of LT/ΔLT OMVs were significantly reduced with inhibition of the JNK pathway. In addition, we noted that JNK inhibition in cells treated with LT+ OMVs and the combination of LT/ΔLT OMVs resulted in similar IL-6 levels (Fig. 4B), suggesting that JNK inhibition removes the contribution of soluble LT to the induction of IL-6 by LT/ΔLT OMVs. It is not clear why the same effect did not occur for TNF-α induction by LT/ΔLT OMVs (Fig. 4A). Together, these results show that the JNK pathway plays no role in LT+ OMV induction of either TNF-α or IL-6 at 6 h and that this contrasts with the significant role that JNK plays in their induction by soluble LT.
AP-1 plays a role in the induction of TNF-α and IL-6 in response to soluble LT but not LT+ OMVs. (A and B) Polarized T84 cells were pretreated for 1 with a JNK inhibitor (10 μM SP600125), followed by incubation with samples standardized to 200 pM LT for 6 h. The levels of TNF-α (A) and IL-6 (B) were measured using RT-PCR. Asterisks indicate significant differences from the corresponding untreated sample: *, P < 0.05; **, P < 0.01; ***, P < 0.005. (C) Polarized T84 cells were incubated with samples standardized to 1 nM LT for 6 h, and nuclear extracts were prepared. Nuclear protein (5 μg) was incubated with a radiolabeled oligonucleotide that corresponded to the DNA-binding region of activated AP-1, and an EMSA was performed to visualize binding. Densitometric values of the shifted DNA band are shown.
To confirm that soluble LT activates the JNK pathway, we performed an electrophoretic mobility shift assay to determine the activation status of AP-1, a downstream effector of JNK. Our results showed that AP-1 was induced in response to both soluble LT and the combination of LT/ΔLT OMVs at 6 h (Fig. 4C), which is consistent with our RT-PCR results. Taken together, our results show that AP-1 plays a role in the induction of IL-6 in response to soluble LT but not OMV-associated LT, providing further evidence that soluble LT and LT+ OMVs act through different mechanisms to induce IL-6 gene expression.
LT and OMV treatments signal differentially through NF-κB.To examine the role of NF-κB in the induction of TNF-α and IL-6 by LT treatments, we inhibited the activation of NF-κB using PDTC, which prevents the binding of NF-κB to DNA. NF-κB inhibition resulted in the downregulation of TNF-α for all samples that contained catalytically active LT (Fig. 5 A). However, PDTC differentially affected IL-6 induction. NF-κB inhibition significantly decreased IL-6 induction by LT+ OMVs, ΔLT OMVs, and S63K OMVs to basal levels, but NF-κB inhibition did not reduce IL-6 gene induction in response to soluble LT (Fig. 5B). Furthermore, we noted a PDTC-dependent decrease in the levels of IL-6 induced by the combination of soluble LT/ΔLT OMVs to a level comparable to that with soluble LT. Thus, NF-κB inhibition appeared to act on the OMV contribution toward IL-6 induction but had no effect on the contribution of soluble LT. These results further support our findings that soluble LT and a comparable concentration of OMV-bound LT induce IL-6 through different pathways.
OMVs but not soluble LT induce IL-6 through NF-κB. Polarized T84 cells were pretreated for 30 min with an NF-κB inhibitor (100 μM PDTC), followed by incubation with samples standardized to 200 pM LT for 6 h. The levels of TNF-α (A) and IL-6 (B) were measured using RT-PCR. Asterisks indicate significant differences from the corresponding untreated sample: *, P < 0.05; **, P < 0.01; ***, P < 0.005.
DISCUSSION
Because ETEC is an important agent of disease worldwide, studies to elucidate the mechanisms of its virulence factors, including LT, are important. Most studies on LT have been performed with its soluble form in many different cell lines. However, recent studies have demonstrated that LT has the propensity to bind LPS and thus become associated with secreted OMVs (17, 18, 31). In addition, LT acts in the gut lumen, where tight barriers generated by the epithelium allow only apical exposure of cells to toxin. Therefore, in this study, we determined the response of polarized intestinal epithelial cells to LT presented in soluble and insoluble contexts.
We found significant differences in the mechanisms and kinetics of TNF-α and IL-6 gene induction elicited by soluble LT and LT+ OMVs. As summarized in Fig. 6, soluble LT elicited IL-6 through two pathways, PKA and JNK. At early time points, in contrast, LT in the context of LT+ OMVs induced IL-6 only through an independent pathway involving NF-κB. Although both OMV-bound and soluble LT elicited TNF-α through some shared pathways, the PKA and JNK pathways were unique to soluble LT. At later times, both LT and LT+ OMVs acted through the PKA pathway (data not shown).
Overview of the different pathways through which soluble LT and LT+ OMVs elicit TNF-α and IL-6 responses in human intestinal epithelial cells. See Discussion for details.
Whereas treatments containing catalytically active LT (soluble LT, LT+ OMVs, and a combination of LT/ΔLT OMVs) elicited substantially higher levels of TNF-α and IL-6 expression at 6 h than non-catalytically active treatments, this higher induction did not depend on the activation of CREB. This result was unexpected. LT acts through the ADP-ribosylation of the Gsα subunit of adenylate cyclase, which results in an increase in cAMP levels. cAMP then activates PKA, which leads to Cl− efflux and the phosphorylation of CREB. As shown in Fig. 2, LT+ OMVs showed delayed kinetics of CREB phosphorylation and CRE gene activity compared to soluble LT and the combination of LT/ΔLT OMVs. Our data suggest that delayed kinetics rather than decreased induction occurs for OMV-associated LT activation, because at 9 h, LT+ OMVs elicited significantly higher levels of IL-6 than soluble LT (P < 0.001), and this increase was significantly inhibited by Rp-camps (data not shown). Because PKA activation is also responsible for phosphorylating the CFTR, resulting in Cl− and water efflux into the intestinal lumen, our results also suggest that the onset of diarrhea may also be delayed in response to LT+ OMVs compared to soluble toxin.
There are many reasons why LT in its native OMV presentation might elicit different responses than soluble LT. Multiple LT molecules are complexed with an OMV, and thus fewer host cells may become intoxicated by OMV-bound LT than by soluble LT. By containing multiple LT molecules, a few LT+ OMVs could elicit maximal CREB activation, whereas soluble LT would elicit a more gradual, proportional response depending on how many individual LT molecules were encountered by each cell. In addition, OMV internalization and trafficking could occur via a different pathway than with soluble LT, as supported by preliminary experiments (data not shown), leading to differences in the efficiency of the intracellular processing of LT. The delayed kinetics of activation could also suggest the inaccessibility of the toxin bound to OMVs. Removal of the LT bound to LPS on the OMVs may be necessary to allow LT to progress through the canonical retrograde trafficking pathway and may be inefficient inside the host cell. Further studies to elucidate the mechanistic basis for the observed differences are ongoing.
H89, which inhibits CREB phosphorylation by PKA at low concentrations, was used as an independent method to demonstrate the role of CREB in the induction of IL-6 and TNF-α. H89 showed results similar to those with Rp-camps, which inhibits PKA activity, except that it led to a significant decrease in TNF-α induction in response to LT+ OMVs. However, this decrease may not be dependent on CREB. In addition to its role as a CREB inhibitor, H89 has also been shown to play a role in the inhibition of mitogen stress kinase 1 (MSK1) (16, 48), which phosphorylates NF-κB to induce NF-κB-mediated gene transcription (48). To test whether this effect was relevant, we tested but did not find any indication of MSK1 phosphorylation or NF-κB phosphorylation in any of our samples using immunoblotting (data not shown). This suggests that MSK1 does not play a role in the induction of TNF-α by LT+ OMVs. H89 has also been shown to inhibit protein kinase D (PKD), which plays a role in Golgi-to-ER transport (27, 35). After binding to the host receptor ganglioside GM1, LT is internalized and transported to the Golgi apparatus and the ER, in which the catalytic subunit is processed. H89 may interfere with this activation by inhibiting the translocation of cargo from the Golgi apparatus to the ER, and this inhibition may also play a role in inhibiting the TNF-α response to LT+ OMVs. This possibility warrants further investigation.
This is the first study to show a role for AP-1 in the host response to LT. We discovered that AP-1 played a role only in the induction of IL-6 by soluble LT and did not affect OMV responses, even at later times. In addition, AP-1 was not activated in response to LT+ OMVs, even at later time points. No previous studies have shown activation of AP-1 by LT, although IL-6 induction in response to increased cAMP levels has been shown to be mediated, at least in part, by AP-1 (7). Dendorfer et al. also showed that cAMP induced IL-6 through mechanisms other than LPS, and the IL-6 response to LPS was completely abrogated by mutations in NF-κB-binding sites (7). We found significant differences not only between soluble LT and LT+ OMVs but also among overall OMV responses. A significant inhibition of IL-6 responses in response to NF-κB inhibition was observed only for treatments containing OMVs. In fact, in the combination of LT/ΔLT OMVs, NF-κB inhibition appeared to remove the OMV contribution to IL-6, resulting in an IL-6 level similar to that induced by soluble LT alone. These results suggest that an OMV component is responsible for activating NF-κB to induce IL-6. Although NF-κB is an important contributor to IL-6 induction, the IL-6 promoter contains multiple regulatory elements, including binding sites for AP-1, NF-κB, and NF-IL-6 (28). Additionally, previous studies have shown that NF-κB is not required for IL-6 induction and that this induction can be mediated by MAP kinases (25, 34). Taken together, our results emphasize the different mechanisms through which soluble LT and OMV treatments induce TNF-α and IL-6 responses.
OMVs were standardized according to lipid content, in consideration of the fact that LPS would probably cause a dominant cytokine response. As a consequence of different protein/lipid ratios in OMVs, however, this normalization process resulted in the protein concentrations of LT+ OMVs and S63K OMVs being more than twice as high as that in ΔLT OMVs. This could have caused the differences in the levels of TNF-α and IL-6 induction. However, we propose that differences in OMV protein concentration were not a major factor in the observed responses. First, despite similar levels of protein in LT+ OMVs and S63K OMVs, the induction of cytokines in response to LT+ OMVs was significantly higher than that in response to S63K OMVs. Second, S63K OMVs, with over 2-fold-higher protein levels, produced levels of TNF-α and IL-6 similar to those produced by ΔLT OMVs. Third, we used samples standardized to two concentrations of LT, 1 nM and 200 pM. Although the amount of protein in the ΔLT OMVs at the 1 nM treatment was twice that in LT+ OMVs at 200 pM, LT+ OMVs induced significantly more IL-6 than ΔLT OMVs.
Our results suggest that the response to the combination of LT/ΔLT OMVs is not merely additive. IL-6 and TNF-α responses induced by the combination of LT/ΔLT OMVs were not equivalent to the combined responses of soluble LT and ΔLT OMVs. These results may be due to the fact that the combination of LT/ΔLT OMVs actually consists of three distinct populations: soluble LT, ΔLT OMVs, and LT bound to the surface of ΔLT OMVs. Although LT may bind to the surface of ΔLT OMVs, this is not the native presentation, because native LT+ OMVs also contain LT in the lumen of the vesicle. The proportion of these populations, and therefore their individual contribution, is difficult to determine. Nevertheless, it was valuable to use this treatment as it allowed us to determine that the context of toxin presentation (i.e., both in the lumen and on the surface of OMVs) was important in eliciting TNF-α and IL-6 responses, which were not merely a result of the presence of equivalent amounts of soluble LT and ΔLT OMVs.
IL-6 and TNF-α are proinflammatory cytokines that mediate acute inflammation. In addition to their role as proinflammatory mediators, IL-6 and TNF-α may also play protective roles in the intestine, such as in tissue repair after injury and protection from apoptosis (38, 47). Because ETEC does not cause disease in mice, mouse models to elucidate the in vivo inflammatory effects of ETEC are not available. In addition, little data have been published on the cytokine responses of patients with diarrhea caused by ETEC. However, other enteric pathogens have been shown to elicit TNF-α and IL-6 responses. Using in vivo studies, Dann et al. showed that IL-6 was produced in mouse intestines in response to Citrobacter rodentium and that this induction was important in preventing infection-induced apoptosis in the colonic epithelium (6). In clinical settings, both IL-6 and TNF-α have been found in the stools of children who had diarrhea that was caused by another enteric pathogen, Shigella dysenteriae (8). In addition, in children with enterocolitis, serum IL-6 has been shown to be discriminative of bacterial etiology from viral etiology (51). Although the mechanisms of IL-6 and TNF-α induction in these studies have not been elucidated, these reports show the relevance of IL-6 and TNF-α induction by enteric pathogens.
In summary, we show that OMV-bound LT is not as effective at inducing CREB activation at early times as soluble LT; however, LT+ OMVs can induce similar amounts of cytokine gene expression. These kinetic differences in CREB activation may be due to different trafficking mechanisms within the cell. Whereas previous studies to determine the effects of LT and vaccination strategies for ETEC have focused on soluble LT, our study emphasizes the importance of studying virulence factors in their native context.
ACKNOWLEDGMENTS
We thank James Fleckenstein for strain H10407 ΔeltA, Stefanie Hartman Chen for assistance with EMSAs, Ben Mudrak for cloning help and critical reading of the manuscript, Patrick Seed for providing the anti-H7 antibody, and Hiroaki Matsunami, Kyla Selvig, and Dan Rodriguez for piloting the CRE-luciferase assay.
This project was supported by grant R01AI064464 from the National Institutes of Health.
FOOTNOTES
- Received 5 May 2011.
- Returned for modification 20 May 2011.
- Accepted 17 June 2011.
- Accepted manuscript posted online 27 June 2011.
- Copyright © 2011, American Society for Microbiology. All Rights Reserved.
