Clostridium difficile causes an intense inflammatory colitis through the actions of two large exotoxins, toxin A and toxin B. IL-8 is believed to play an important role in the pathophysiology of C. difficile-mediated colitis, although the mechanism whereby the toxins up-regulate the release of IL-8 from target cells is not well understood. In this study, we investigated the mechanisms through which toxin A induces IL-8 secretion in human monocytes. We found that cellular uptake of toxin A is required for the up-regulation of IL-8, an effect that is not duplicated by a recombinant toxin fragment comprising the cell-binding domain alone. Toxin A induced IL-8 expression at the level of gene transcription and this effect occurred through a mechanism requiring intracellular calcium and calmodulin activation. Additionally, the effects of toxin A were inhibited by the protein tyrosine kinase inhibitor genistein, but were unaffected by inhibitors of protein kinase C and phosphatidylinositol-3 kinase. We determined that toxin A activates nuclear translocation of the transcription factors NF-κB and AP-1, but not NF-IL-6. NF-κB inhibitors blocked the ability of toxin A to induce IL-8 secretion, and supershift analysis indicated that the major isoform of NF-κB activated by the toxin is a p50-p65 heterodimer. This study is the first to identify intracellular signaling pathways and transcription factors involved in the C. difficile toxin-mediated up-regulation of IL-8 synthesis and release by target cells. This information should increase our understanding of the pathogenesis of C. difficile colitis and the nature of IL-8 gene regulation as well.
Clostridium difficile is an important emerging bacterial pathogen and a leading cause of nosocomial diarrhea in the United States (1, 2, 3). The spectrum of C. difficile-associated disease (CDAD)3 ranges from a mild antibiotic-associated diarrhea, to severe, or even life-threatening, pseudomembranous colitis. CDAD is caused by the actions of two large exotoxins, toxin A and toxin B, which are produced by toxigenic strains of C. difficile (4, 5). Toxins A and B possess a high degree of structural and functional homology (6, 7, 8, 9, 10) and exhibit a three-domain structure characteristic of other large clostridial toxins (11). The carboxyl-terminal portion of the toxins comprises the cellular-binding domain, the middle third includes a hydrophobic element, and the amino-terminal portion contains the enzymatic activity (UDP-glucosyltransferase). Toxins A and B are able to transfer the glucose moiety of cellular UDP-glucose to certain intracellular signaling proteins known as the Rho family of small GTPases (9, 10, 12). Toxin-mediated glucosylation inactivates the Rho GTPases and leads to disruption of cytoskeletal integrity and cytotoxic effects. In addition to glucosyltransferase activity, both toxins induce chemokine secretion from certain cells in vitro (13, 14, 15, 16, 17). Furthermore, toxin A, but not toxin B, can produce intense fluid accumulation, inflammation, secretion of chemokines such as IL-8, and tissue destruction in ligated sections of lamb and rabbit ileal loops (6, 18).
An intense inflammatory response with a marked neutrophil accumulation is a key characteristic of the clinical pathophysiology of CDAD (3, 19). Although inflammation is an important arm of the immune system and represents an early line of defense against many potential pathogens, unregulated inflammation can contribute to the destruction of local host tissues, a phenomenon that may be relevant to certain aspects of CDAD. The chemokine IL-8 plays an important role in the pathogenesis of neutrophilic types of inflammation. Indeed, stools and colonic mucosal biopsies from patients with C. difficile colitis contain elevated levels of IL-8, implicating a role for this chemokine in the pathophysiology of CDAD (20). IL-8 is a CXC-type chemokine that binds to the cellular seven-transmembrane domain G protein-coupled receptors known as CXCR1 and CXCR2 (21). IL-8 is a potent proinflammatory chemotactic factor that predominantly exerts its effects on neutrophils (22, 23). Because IL-8 is such a powerful cytokine, its extracellular release must be tightly regulated. In most cell types, transcriptional control of IL-8 expression appears to be the most important regulatory mechanism (24, 25, 26). The proximal 135 bp of the 5′-flanking region of the IL-8 gene contains four elements known to be involved in transcriptional regulation: AP-1, CCAAT/enhancer-binding protein (C/EBP/NF-IL-6) and NF-κB sites up-regulate transcription, and an Octamer-1 (Oct-1) element that can repress transcription of the gene (26, 27).
NF-κB was originally identified as the transcription factor that activates the promoter of the κ region of the Ig light chain locus in B lymphocytes (28). Since its initial discovery, investigators have demonstrated its involvement in the positive regulation of a number of other genes including those for IL-8, IL-2, IL-6, monocyte chemoattractant protein-1, G-CSF, TNF-α, the enzymes inducible nitric oxide synthase and cyclooxygenase-2, the adhesion molecules ICAM-1 and E-selectin, and its own repressor, IκBα (28). NF-κB exists as a family of homodimers and heterodimers composed of different combinations of proteins belonging to the NF-κB/Rel family (29). In most cells, NF-κB is normally complexed with its inhibitor protein, IκB. IκB constrains NF-κB to the cytoplasm by masking its nuclear localization signal (30, 31). Active NF-κB is transported to the nucleus after IκB is phosphorylated by IκB-kinase and degraded by the proteasome complex (32).
C. difficile toxin A induces IL-8 secretion from a large variety of cell types, including peripheral blood monocytes and the monocyte cell line THP-1 (15), as well as the intestinal epithelial cell lines HT-29 and T-84 (14, 33). Because the mechanism through which toxin A induces IL-8 secretion is unknown, we sought to characterize the nature of IL-8 up-regulation induced by toxin A in human monocytes. In this study, we found that the induction of IL-8 synthesis and secretion requires cellular uptake of toxin A and cannot be stimulated by a recombinant toxin fragment containing only the cell-binding domain. We also determined that toxin A induces IL-8 expression at the level of gene transcription and that this occurs through a mechanism that requires intracellular calcium and likely involves a calmodulin-sensitive step. Toxin A-induced IL-8 secretion was partially attenuated by the tyrosine kinase inhibitor genistein, but was unaffected by inhibitors of protein kinase C (PKC) and phosphatidylinositol-3 kinase (PI-3 kinase). In addition, we determined that toxin A activates nuclear translocation of NF-κB and AP-1. Pharmacological inhibition of the activation of NF-κB inhibited the ability of toxin A to induce IL-8 secretion, suggesting an important role for NF-κB in this process. Supershift experiments indicated that the isoform of NF-κB activated by toxin A is the p50-p65 heterodimer. These findings characterize an integral part of the mechanism of toxin A-induced IL-8 secretion and serve as a model to further our knowledge of the inflammatory response to enteric pathogens. Understanding the mechanism through which toxin A induces a pathological inflammatory response in CDAD may help lead to the development of a novel nonantibiotic approach to therapy of this disease.
Materials and Methods
Human PBMC were isolated from whole blood or buffy coat fractions obtained from normal volunteers by density centrifugation using Ficoll Histopaque 1077 (Sigma, St. Louis, MO). The PBMC were enriched for monocytes by adherence to plastic tissue culture dishes for 90 min at 37°C and used predominantly for the IL-8 assays (according to the method of Linevsky et al.) (15). Based on Giemsa staining, >80% of cells purified by this technique were determined to be monocytes. For experiments requiring large numbers of cells (e.g., gel-shift and supershift assays), monocytes were isolated from PBMC using a CD-14 magnetic bead column-based, positive cell selection technique (Miltenyi Biotec, Auburn, CA) and were determined to be >90% monocytes. The two techniques for monocyte isolation yielded identical results with respect to gel-shift and IL-8 assays. Monocytes were allowed to rest in RPMI 1640 (Life Technologies, Grand Island, NY) supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin overnight before use. The discarded nonadherent or column pass-through cells were predominantly lymphocytes and these cellular fractions exhibited negligible or nondetectable IL-8 release and activation of NF-κB in response to toxin A (data not shown).
Purified toxin A and TcdA.3 were gifts from D. Lyerly and T. Wilkins (Techlab, Blacksburg, VA). Several different lots of purified toxin A were used. Each was found to be free of LPS contamination and of equivalent activity. TcdA.3 is a recombinant GST-fusion polypeptide comprising the cell-binding domain of toxin A. GST alone did not have any effect on IL-8 production or EMSA results (data not shown). Because TcdA.3 is prepared from Escherichia coli, the preparations contained variable amounts of LPS as assayed by the E-TOXATE Limulus test (Sigma). For our studies, a polymyxin B agarose column (Sigma) was used to remove LPS from the TcdA.3 preparations prior to use. Different lots of TcdA.3 produced similar results.
Monocytes at 106 cells/ml were stimulated with PBS alone or containing various concentrations of toxin A or TcdA.3 for 4 h at 37°C. In some experiments the cells were preincubated with vehicle control or various inhibitors for 30 min prior to the addition of toxin A. Compounds from Calbiochem (San Diego, CA) were used at the following final concentrations: EDTA (2 mM), 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetra(acetoxymethyl) ester (BAPTA/AM) (30 μM), chloroquine (0.1 mM), NH4Cl (20 mM), sulfasalazine (2 mM), curcumin (20 μM), wortmannin (100 nM), calpastatin (200 nM), genistein (26 μM), and bisindolylmaleimide (100 nM). The inhibitor W-7 (N-(6-aminohexyl)-5-chloro-1-napthalenesulfonamide hydrochloride) was from Sigma and was used at a final concentration of 20 μM. For experiments using immobilized toxins, 10 μg of LPS, toxin A, or TcdA.3 in 50 μl of binding buffer (1.5% NaHCO3, pH 9.6) was added to 96-well Immulon IV plates (Dynex, Chantilly, VA) overnight at 37°C, as previously described (34). The wells were then washed twice with PBS to remove unbound protein, and monocytes were incubated in the wells for 4 h. Cell supernatants were analyzed for IL-8 using the Quantikine ELISA for human IL-8 (R&D Systems, Minneapolis, MN) or the IL-8 EASIA (BioSource International, Camarillo, CA); results were equivalent using either kit. The values displayed in Figs. 1 A, 2, 4, 5, 6, and 7 represent the mean and SDs from representative experiments as described in the figure legends.
Detection of IL-8 mRNA
Unfractionated RNA was purified from monocytes and was analyzed by slot blotting and Northern hybridization essentially as described previously (35). A specific 30-base ssDNA oligonucleotide (Operon, Alameda, CA) corresponding to a region in the 5′-untranslated region of IL-8 mRNA was end-labeled with [γ-32P]ATP (ICN, Costa Mesa, CA). Using the conditions outlined here, this probe recognizes a single band corresponding to IL-8 mRNA in fractionated cellular RNA (Ref. 35 , and data not shown). Briefly, total mRNA was extracted from monocytes using RNA STAT-60 (Tel-Test, Friendswood, TX), and 15 μg of RNA was immobilized on a Nytran+ (Schleicher & Schuell, Keene, NH) membrane using a slot-blot apparatus (Life Technologies). The blot was hybridized overnight at 60°C, washed in 1× SSC, 0.1% SDS, and autoradiographed at −80°C for 24 h. The results were quantified by densitometric analysis (NIH-Image computer program) (36). Fig. 1 B displays results representative of three separate experiments.
Nuclear extracts were prepared essentially as previously described by Smith et al. (37). Monocytes at 107 cells/ml were stimulated with 10 μg/ml LPS or various concentrations of toxin A for 30 min at 37°C. A time course indicated that maximal NF-κB activity was observed at 30 min after the addition of toxin A (data not shown). In some experiments toxin A was preincubated with a neutralizing toxin A affinity-purified polyclonal goat antiserum (a gift from T. Wilkins, VPI, Blacksburg, VA) for 30 min prior to addition of the toxin to monocytes (34). In other experiments, monocytes were preincubated with the NF-κB inhibitors sulfasalazine (2 mM) or curcumin (20 μM) for 30 min prior to toxin A treatment. Nuclear extracts were prepared as previously described and either used immediately for EMSA or stored at −80°C for future use (37).
EMSA and supershift assays
EMSA were performed essentially as described by Smith et al. (37). Consensus NF-κB or AP-1 oligonucleotide probes were from Promega (Madison, WI) and the NF-IL-6 probe 5′-TCGATCAGTTGCAAATCGT-3′ was custom synthesized (IDT, Coralville, IA). All probes were end-labeled with [γ-32P]ATP (ICN). Reaction mixtures consisted of 2 μg of nuclear extract, binding buffer (20 mM HEPES, 1 mM DTT, 0.1 mM EDTA, 50 mM KCl, 5% glycerol, 200 μg/ml BSA), 1 μg poly(dI-dC) (Sigma), and 0.05 μg of sonicated salmon sperm (Stratagene, La Jolla, CA). Anti-Rel Abs for supershifts were purchased from Santa Cruz (Santa Cruz, CA): α-p50 (sc-114), α-p65 (sc-109), α-cRel (sc-70), and α-RelB (sc-226); 2 μl were added per sample. Ten nanograms of cold NF-κB or AP-1 consensus probe were added to the control samples as specific and nonspecific competitors. The reactions were incubated on ice for 30 min, 1 ng of γ32P-labeled probe was added, incubated for 10 min at room temperature, and immediately subjected to nondenaturing 4% PAGE. The gels were dried and autoradiographed at −80°C for 18 h. The images were analyzed semiquantitatively by densitometry using the NIH-Image computer program. Figs. 8, 9, 10, and 11 display representative experiments as described in the figure legends.
Toxin A-induced secretion of IL-8 from human monocytes results from transcriptional up-regulation of the IL-8 gene
Linevsky et al. previously demonstrated by RT-PCR that toxin A up-regulates IL-8 gene expression (15). We sought to confirm and extend this finding and to determine whether or not transcriptional up-regulation of the IL-8 gene is the predominant mechanism through which toxin A induces IL-8 secretion in monocytes. Pretreatment of monocytes with the transcriptional inhibitor actinomycin D completely abolished the ability of toxin A to induce IL-8 secretion (Fig. 1,A), indicating that IL-8 secretion is dependent upon transcription. To further characterize the nature of the IL-8 response, we performed Northern analysis on RNA isolated from monocytes treated with various concentrations of toxin A. Minimal IL-8 message was detected in resting monocytes. Toxin A treatment of monocytes, however, stimulated a marked, dose-dependent increase in detectable IL-8 mRNA (Fig. 1 B). By semiquantitative densitometric analysis we determined that 0.1 nM toxin A resulted in an ∼13-fold increase in IL-8 message, and 1.0 nM and 10 nM toxin A induced ∼19- and ∼30-fold increases, respectively. Overall, these results indicate that toxin A up-regulates IL-8 secretion primarily through transcriptional up-regulation of the IL-8 gene in monocytes.
Toxin A-induced IL-8 secretion requires mobilization of intracellular Ca2+ and activation of calmodulin
Excluding effects on Rho GTPases, little is known about the intracellular signaling pathways induced or altered by toxin A. Toxin A has been reported to induce release of Ca2+ from intracellular stores in neutrophils (38, 39) and rat pancreatic acini (40). Because Ca2+ ionophores have been shown to elicit IL-8 secretion from monocytes (41), and an intracellular Ca2+ flux is associated with induction of IL-8 secretion by certain signals, such as IL-1 (42), we next sought to determine whether the effect of toxin A on IL-8 secretion required mobilization of intracellular Ca2+. We used EDTA to deplete extracellular Ca2+ and used the cell-permeable agent BAPTA/AM to chelate releasable intracellular stores of Ca2+. Monocytes incubated in EDTA alone showed a decrease in the level of toxin A-induced IL-8 secretion by approximately 40% compared with control monocytes (Fig. 2,A). When monocytes were incubated in EDTA and BAPTA/AM, the ability of toxin A to activate IL-8 secretion was completely inhibited (Fig. 2,A). Because a major effect of intracellular Ca2+ flux is the activation of calmodulin, we next examined whether the inhibition of calmodulin could diminish toxin A-induced secretion of IL-8. W-7 is a potent and specific inhibitor of calmodulin activity and we observed that treatment with this agent abrogated toxin-mediated IL-8 secretion by ∼83% (Fig. 2 B). These results indicate that both intracellular Ca2+ and calmodulin are required for toxin A-induced IL-8 secretion in monocytes.
TcdA.3 is not sufficient to induce of IL-8 secretion in monocytes
Toxin A binds to target cells through its carboxyl-terminal repeating domain (Fig. 3), although the identity of the human cell surface receptor is not yet known (43). A recombinant GST-fusion polypeptide comprising this cell-binding domain of toxin A has been reported to induce intracellular Ca2+ flux in rabbit brush border epithelial cells (43). Using this information in conjunction with our observation that the toxin A-induced up-regulation of IL-8 secretion contained a Ca2+-sensitive step, we next sought to determine whether treatment of monocytes with the cell-binding domain of toxin A alone (referred to as TcdA.3) could induce IL-8 secretion. Monocytes were treated with either native holotoxin A or recombinant TcdA.3 for 4 h. In agreement with a previous report (15), we found that toxin A-induced IL-8 secretion is dose-dependent with an optimal dose of 10−9 M (Fig. 4). Somewhat unexpectedly, however, we found that the TcdA.3 fragment did not induce appreciable levels of IL-8 production from monocytes (Fig. 4). Even after 24 h, TcdA.3 did not induce significant levels of IL-8 release (data not shown).
Internalization of toxin A is required for IL-8 secretion
In order to exert their cytotoxic effects, toxins A and B must be internalized by the target cell (44, 45). Lysosomotropic agents such as chloroquine and NH4Cl can inhibit toxin-mediated cytotoxicity, suggesting that endosomal acidification is involved in internalization (45). In order to determine whether internalization of toxin A is also required for the toxin-mediated induction of IL-8 secretion, we pretreated monocytes with these agents prior to the addition of toxin A and observed that both chloroquine and NH4Cl effectively inhibited toxin A-induced secretion of IL-8 (Fig. 5,A). We assessed the importance of toxin internalization, using an additional approach. Tissue culture plates were coated with toxin A or TcdA.3 to supply the stimulus for IL-8 release. Monocytes that attach and spread on this solid phase support containing toxin A would be able to bind, but not internalize toxin A (46). As a positive control, we observed that monocytes allowed to adhere to LPS-coated wells secreted significant amounts of IL-8 (Fig. 5 B). Monocytes adherent to toxin A-coated wells, however, failed to secrete detectable IL-8. Taken together these results strongly suggest that cellular uptake and internalization through an endosome-like pathway is required for toxin A to effectively induce IL-8 secretion from monocytes.
Effects of kinase inhibitors on toxin A-mediated IL-8 secretion
The reversible phosphorylation of serine, threonine, and tyrosine residues acts as a molecular switch to translate signals from the extracellular environment into specific cell functions. Protein kinase C (PKC) is a serine/threonine kinase that has been shown to play a role in the induction of IL-8 secretion by Helicobacter pylori, PMA, and IL-1β (47, 48, 49). In addition, intracellular Ca2+ flux activates “classical” forms of PKC. However, we found that treatment of monocytes with the PKC inhibitor bisindolylmaleimide did not diminish the effect of toxin A and, in fact, actually caused a slight increase in basal and toxin-induced levels of IL-8 release (Fig. 6). Induction of IL-8 secretion by bisindolylmaleimide has been observed in human synovial fibroblasts (50). Similarly, although phosphatidylinositol-3 kinase (PI-3 kinase) is known to be an important effector of a variety of intracellular signaling pathways in monocytes, the PI-3 kinase inhibitor wortmannin did not block the up-regulation of IL-8 by toxin A and seemed to cause an increase in the basal level of IL-8 secretion (Fig. 6). IL-8 production in response to certain agents, such as IL-1β, has been reported to involve protein tyrosine kinase (PTK) activity (49). When used in our system, we observed that the tyrosine kinase inhibitor genistein did decrease toxin A-induced release of IL-8 by about 75% (Fig. 6), indicating a potential role for tyrosine kinase-dependent signaling pathways in the up-regulation of IL-8 release stimulated by toxin A.
Inhibitors of the transcription factor NF-κB abrogate the toxin A-mediated activation of IL-8 secretion
NF-κB is a potent transcriptional activator of various inflammatory mediators. For many types of stimuli, including certain microbial pathogens, NF-κB appears to be required for inducible transcription of the IL-8 gene (51, 52). Therefore, we next examined whether there was evidence for a role of NF-κB in toxin A-induced IL-8 secretion. Sulfasalazine and curcumin are well-described inhibitors of NF-κB activity in a variety of cell types. Each agent is believed to act by a different mechanism: sulfasalazine inhibits phosphorylation of IκBα (53), whereas curcumin interacts directly with the p50 subunit of NF-κB, thereby blocking degradation of IκBα (54). Pretreating monocytes with these agents, we observed that sulfasalazine completely abolished the ability of toxin A to induce IL-8 release and curcumin markedly diminished (by ∼75%) the effect of toxin A as well (Fig. 7). These results suggest that activation of NF-κB is required for the toxin-induced up-regulation of IL-8 expression.
Toxin A induces nuclear translocation of NF-κB and AP-1 but not NF-IL-6
Based on our results using the NF-κB inhibitors, it seemed likely that toxin A would be able to activate and induce nuclear translocation of NF-κB. Activation and nuclear translocation of NF-κB can be detected with EMSA. Using a NF-κB-specific probe to perform EMSA we found that resting monocytes contain a low level of nuclear NF-κB (Fig. 8, lane 1). Within 30 min of stimulation with 1 nM toxin A, we observed a roughly sixfold increase in the amount of nuclear NF-κB (lane 2). LPS induces a similar increase in the amount of nuclear NF-κB (lane 3). We confirmed the specificity of the complex that was induced by toxin A through competition analysis; a 10-fold excess of unlabeled NF-κB oligonucleotide resulted in a reduction of the predominant complex induced by toxin A (lane 4), whereas a 10-fold excess of an unrelated oligonucleotide (AP-1) failed to compete (lane 5). The toxin A-induced NF-κB activation occurred in a concentration-dependent manner and could be blocked by preincubation with a specific antitoxin A polyclonal Ab (data not shown).
The IL-8 promoter contains three elements known to be involved in transcriptional up-regulation. The NF-κB element is believed to be the main regulator of inducible expression of the IL-8 gene; however, for certain signals, expression of the IL-8 gene has been reported to depend upon activation of a second transcription factor, either AP-1 or NF-IL-6, in addition to NF-κB (51, 52). For example, human cytomegalovirus-induced transcriptional up-regulation of the IL-8 gene is dependent upon both AP-1 and NF-κB (51). We therefore investigated the effect of toxin A on AP-1 and NF-IL-6. Using an AP-1-specific probe, we found that the basal level of AP-1 in monocytes was fairly high (Fig. 9, lane 1), but we observed a modest, approximately twofold, increase in the amount of nuclear AP-1 in toxin-treated nuclear extracts (lane 2). We confirmed the specificity of the complex detected by EMSA through competition analysis; a 10-fold excess of unlabeled AP-1 oligonucleotide resulted in a reduction of the complex induced by toxin A (lane 3), whereas a 10-fold excess of an unrelated oligonucleotide (NF-κB) failed to compete (lane 4). In contrast to the results observed for NF-κB and AP-1, we detected no significant increase in nuclear localization of the transcription factor NF-IL-6 in response to toxin A (data not shown).
Based on the results observed in the IL-8 experiments summarized in Fig. 7, we predicted that pharmacological inhibitors of NF-κB should diminish the effect of toxin A on activation of NF-κB as well. Indeed, we found that sulfasalazine (Fig. 10, lane 4) and curcumin (lane 6) were each able to markedly inhibit toxin A-induced nuclear localization of NF-κB. Overall, then, our results indicate a central role for NF-κB, likely in conjunction with AP-1, in the up-regulation of IL-8 gene expression induced by toxin A.
Toxin A activates the p50/p65 NF-κB heterodimer
NF-κB is actually a complex of homo- or heterodimers comprised of members of the Rel family of proteins. The p50/p50 NF-κB homodimer lacks a transcriptional activation domain and, in some instances, acts as a transcriptional repressor (55, 56). The p65 Rel subunit possesses a transcriptional-activating domain, hence the p50/p65 form of NF-κB usually functions as a powerful inducible transcriptional activator (55). In order to identify the subunits present in the toxin A-activated isoforms of NF-κB, we performed supershift assays using Abs specific for each of the known Rel proteins (Fig. 11). Abs to p50 (lane 5) and p65 (lane 6) induced a partial shift in mobility, and a partial loss of the predominant toxin-inducible NF-κB complexes. When the p50 and p65 Abs were added to extracts simultaneously, the toxin-induced NF-κB complex was almost completely shifted (lane 7). Abs to c-Rel (lane 8) and Rel-B (lane 9) did not induce a detectable shift or loss of the toxin-induced NF-κB complex. This result suggests that toxin A induces nuclear localization of the p50/p65 heterodimer isoform of NF-κB.
Inflammation and neutrophil infiltration are central to the pathophysiology of CDAD. Neutrophilic infiltration can be so pronounced in patients with CDAD that the neutrophils, together with fibrin, form macroscopic plaques, known as pseudomembranes, on the colonic wall (1). Release of reactive oxygen species and enzymes from activated neutrophils likely acts synergistically with toxins A and B to produce tissue destruction in CDAD. Studies suggest that IL-8 is the principle cytokine involved in migration and activation of neutrophils and, in fact, elevated levels of IL-8 have been detected in stool specimens from patients with CDAD (20, 57). Of the two C. difficile toxins, the main effector of IL-8 secretion in CDAD appears to be toxin A, which is able to induce neutrophilic inflammation in rabbit ileal loops as well as IL-8 secretion from colonic epithelial cells and monocytes in vitro (14, 15, 58). However, despite the evidence of its important role in the pathophysiology of CDAD, the intracellular mechanisms leading to toxin A-induced IL-8 secretion have not been identified. The goal of this investigation was to characterize the biomolecular mechanism of toxin A-induced IL-8 secretion. We found, in agreement with a report from Linevsky (15), that 1 nM toxin A induces nearly a 100-fold increase in IL-8 secretion from primary human monocytes. In addition, through treatment with actinomycin D and by Northern analyses, we determined that toxin A up-regulates release of IL-8 from monocytes primarily by stimulating transcription of the IL-8 gene.
Intracellular calcium flux is an important second messenger that plays a role in a diverse array of cellular processes, including nuclear signaling. Certain extracellular signals can induce rapid mobilization of intracellular stores and/or promote influx of extracellular Ca2+. Ca2+ pumps revert the level of cytosolic Ca2+ to its normal state and the resulting Ca2+ oscillations produced by this on/off mechanism can impart specificity of the signal as it is translated by the cell (59). Signaling through Ca2+ mobilization has been implicated in the activation of NF-κB as well as the up-regulation of IL-8 by stimuli such as NO and thapsigargin (42, 59, 60). Our findings implicate key roles for intracellular Ca2+ flux and calmodulin in the activation of IL-8 release induced by toxin A. Other investigators have reported mobilization of intracellular Ca2+ by toxin A (38, 39, 40, 43). Our study, however, is the first to link an intracellular Ca2+ signal to any of the known cellular effects of toxin A. It is unlikely that Ca2+ and calmodulin are required for binding or internalization of toxin A because a previous study demonstrated that they are not necessary for toxin A-mediated cytotoxicity in rat intestinal crypt cells (61). In addition toxin-induced cytotoxicity is known to be dependent on cellular internalization of the toxin (44, 45). In our study, complete inhibition of the effects of toxin A required both EDTA and BAPTA/AM, indicating that mobilization of intracellular Ca2+ and influx of extracellular Ca2+ are each required for effective activation of IL-8 gene transcription. In addition, the specific calmodulin inhibitor W-7 abrogated IL-8 secretion, suggesting that the effect of toxin A requires the activation of calmodulin. Together, these results strongly implicate the involvement of an intracellular Ca2+/calmodulin-dependent signaling pathway in toxin A-induced secretion of IL-8. Potential roles for Ca2+ and calmodulin may involve the regulation of endosomal maturation and activation of kinases involved in the pathway NF-κB signaling pathway.
Our study also demonstrates that the carboxyl-terminal repeating portion of toxin A alone (TcdA.3) was not able to induce IL-8 secretion from monocytes. In some ways, this result was unexpected as TcdA.3 has been reported to be sufficient to induce intracellular Ca2+ flux in neutrophils and rabbit brush border epithelial cells (42, 43). One possible explanation is that the toxin A-dependent signaling pathways are different for monocytes as compared with other cell types. Alternatively, the binding portion may actually direct the signaling necessary to induce the IL-8 response, but the recombinant fusion protein, TcdA.3, does not retain the proper signaling conformation. Another possibility is that a signal in addition to Ca2+ mobilization is required to induce IL-8 expression, a signal that is transduced by the holotoxin, but not the TcdA.3 fragment. If this is the case, then some other portion of the toxin molecule may be required to up-regulate expression of IL-8. We also found that internalization and endosomal acidification are necessary prerequisites to toxin-mediated IL-8 secretion. Barroso et al. used deletional analysis of the toxin B gene to determine that a 50-amino acid hydrophobic portion of the middle domain of the toxin is involved in toxin internalization (62). It is therefore possible that TcdA.3 lacks IL-8-inducing activity because it is not properly internalized without the central domain.
Transcriptional activation of cytokine genes such as IL-8 occurs when a stimulus interacts with the cell in such a way that its message is relayed to the transcription factors directly responsible for transcriptional activation of the gene. This signal transduction pathway generally involves the sequential activation of a cascade of intracellular protein kinases. In an attempt to characterize the signaling pathway through which toxin A induces IL-8 secretion in monocytes we employed inhibitors for three important families of kinases: PKC, PI-3 kinase, and protein tyrosine kinase (PTK). Bisindolylmaleimide is a selective inhibitor active against PKC-α, -βI, -βII, -γ, -δ, and -ε isoforms (63). We found that the treatment of monocytes with this agent did not inhibit toxin A-induced IL-8 secretion. In agreement with this finding, Hippensteil et al. reported that toxin B does not activate PKC and may actually inhibit PMA-mediated PKC activity (64). We also found that the PI-3 kinase inhibitor wortmannin was unable to block toxin A-mediated IL-8 secretion. In support of this finding, it has been reported that toxin B does not up-regulate PI-3 kinase in monocytes and actually inhibits LPS-induced PI-3 kinase activation (65). The PTK inhibitor genistein, however, did abrogate toxin A-induced IL-8 secretion in this study. Precedent exists for this finding, as the PTK inhibitor erbistatin was found to inhibit toxin A-induced secretion in rabbit ileal loops (R. L. Guerrant, University of Virginia, personal communication) and PTKs have been implicated in IL-1β- and H. pylori-induced IL-8 secretion (49, 66). This report, however, is the first to implicate a PTK-sensitive pathway for a C. difficile toxin-mediated cellular effect.
Activation of NF-κB by enteric bacterial pathogens such as H. pylori and enteropathogenic E. coli has been reported by others, but our report is the first to demonstrate the activation of this important proinflammatory transcription factor by toxins from C. difficile (67, 68). We found that toxin A induces rapid nuclear translocation of the p50/p65 isoform of NF-κB in monocytes. In addition, our findings that the NF-κB inhibitors sulfasalazine and curcumin abrogate toxin-mediated IL-8 secretion suggests that NF-κB plays a functional role in this process and is not a simple concurrent but unrelated phenomenon. The time course of the toxin-mediated activation of NF-κB in relation to IL-8 secretion also supports this hypothesis. Our data are consistent with the observations that, for most stimuli, the activation of NF-κB is essential for inducible expression of the IL-8 gene and, in some instances, has been noted to occur following Ca2+ flux and preceding IL-8 production (26, 54, 59).
Even though NF-κB appears to be the most important transcription factor for inducible expression of the IL-8 gene, most stimuli must also activate AP-1 or NF-IL-6 to elicit IL-8 expression. Our study demonstrates that AP-1 is also activated by toxin A. Since certain kinases such as MEKK-1 are common to the pathways leading to activation of NF-κB and AP-1, some stimuli including TNF-α can induce both transcription factors (69). Furthermore, several stimuli such as human cytomegalovirus have been shown to induce IL-8 expression through the activation of both NF-κB and AP-1 (51).
In summary, C. difficile toxin A up-regulates IL-8 expression at the transcriptional level and intracellular Ca2+/calmodulin, a PTK, and the transcription factor NF-κB play important roles in this effect. Cell surface binding of toxin A is not sufficient to induce IL-8 secretion; internalization appears to be required. The activation of NF-κB by toxin A may play other roles in addition to the expression of IL-8 because NF-κB is also involved in the transcriptional activation of other cytokines known to be induced by toxin A, including TNF-α, IL-1, and IL-6, (13, 16). Current therapy for CDAD requires treatment with the antibiotic vancomycin or metronidazole although relapse occurs in 20% of cases (70). Because antibiotic therapy inhibits reconstitution of normal colonic flora it may contribute to the high rate of CDAD recurrences. Our finding that NF-κB plays a role in the toxin A-mediated neutrophilic inflammatory response could potentially provide a target for the development of novel nonantibiotic treatment for CDAD. In addition, further investigation of the biochemical pathways leading to the toxin A-mediated up-regulation of IL-8 may uncover novel aspects of the nature of IL-8 gene expression itself.
We thank Drs. David Lyerly, Tracy Wilkins, and Scott Moncrief for their generous gifts of purified toxin A and TcdA.3 and for their helpful discussions. We also thank Dr. William Ciesla, Jr., and other colleagues at the University of Virginia for helpful comments and advice.
This work was supported in part by grants from the National Institutes of Health (R01-GM54572 to D.A.B. and R29-AI34358 (to M.F.S.). K.K.J. is supported by National Institutes of Health Training Grant T32AI07046).
Abbreviations used in this paper: CDAD, C. difficile-associated disease; PKC, protein kinase C; PTK, protein tyrosine kinase; PI-3 kinase, phosphatidylinositol-3 kinase; BAPTA/AM, 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetra(acetoxymethyl) ester; TcdA.3, cell-binding domain of toxin A.