|Home | About | Journals | Submit | Contact Us | Français|
The type II heat-labile enterotoxins (LT-IIa and LT-IIb) of Escherichia coli have an AB5 subunit structure similar to that of cholera toxin (CT) and other type I enterotoxins, despite significant differences in the amino acid sequences of their B subunits and different ganglioside receptor specificities. LT-II holotoxins and their nontoxic B subunits display unique properties as immunological adjuvants distinct from those of CT and its B subunits. In contrast to type II holotoxins, the corresponding pentameric B subunits, LT-IIaB and LT-IIbB, stimulated cytokine release in both human and mouse cells dependent upon Toll-like receptor 2 (TLR2). Induction of interleukin-1β (IL-1β), IL-6, IL-8, or tumor necrosis factor alpha in human THP-1 cells by LT-IIaB or LT-IIbB was inhibited by anti-TLR2 but not by anti-TLR4 antibody. Furthermore, transient expression of TLR1 and TLR2 in human embryonic kidney 293 cells resulted in activation of a nuclear factor-κB-dependent luciferase gene in response to LT-IIaB or LT-IIbB. Moreover, peritoneal macrophages from TLR2-deficient mice failed to respond to LT-IIaB or LT-IIbB, in contrast to wild-type or TLR4-deficient cells. These results demonstrate that besides their established binding to gangliosides, the B subunits of type II enterotoxins also interact with TLR2. Although a ganglioside-nonbinding mutant (T34I) of LT-IIaB effectively induced cytokine release, a phenotypically similar point mutation (T13I) in LT-IIbB abrogated cytokine induction, suggesting a variable requirement for gangliosides as coreceptors in TLR2 agonist activity. TLR2-dependent activation of mononuclear cells by type II enterotoxin B subunits appears to be a novel mechanism whereby these molecules may exert their immunomodulatory and adjuvant activities.
In similarity with cholera toxin (CT) and other type I enterotoxins, the Escherichia coli type II toxins, LT-IIa and LT-IIb, display an AB5-type structure in which the toxic A subunit is noncovalently linked to a pentameric binding (B) subunit (16). However, type I and II enterotoxins share less than 14% amino acid sequence identity in their B subunits and exhibit differential binding to ganglioside receptors (16). Type I toxins bind with high affinity to ganglioside GM1. LT-IIa displays a more promiscuous binding profile, which includes GD1b, GD1a, and GM1, in order of decreasing affinity. The LT-IIb variant lacks affinity for GM1 or GD1b but binds avidly to GD1a (10). Enterotoxins have attracted considerable attention also due to their exceptional adjuvant properties (28). Our recent work has demonstrated that LT-II toxins and nontoxic derivatives thereof possess strong immunoenhancing activity, quite distinct from that of CT and analogous nontoxic structures (20, 21).
Upon binding to gangliosides, enterotoxins cause elevation of intracellular cAMP levels in a variety of eukaryotic cell types (16, 28). In enterocytes, cAMP elevation leads to massive secretion of fluid and electrolytes into the gut lumen, causing watery diarrheal symptoms (16, 28). Since several cytokine genes contain cAMP-responsive elements involved in their transcriptional regulation (2, 27), gangliosides on monocytic cells may function as innate host receptors for enterotoxins. Gangliosides recognize diverse microbial structures and may thus possess pattern recognition capabilities. For example, in addition to the LT-II toxins, GD1a recognizes the Pseudomonas aeruginosa flagellin (6) and the polyoma virus (41). Similarly, GM1 and GD1b also recognize the simian virus 40 (41) and the fragment C of tetanus toxin (30), respectively.
The study of interactions between innate immune cells and type II enterotoxins may reveal mechanisms associated with their immunomodulatory and adjuvant activities. Indeed, it is now well established that activation of innate immunity through pattern recognition of microbial structures is essential not only in first-line defense but also in initiating the adaptive immune response (36). In this regard, the cooperation between certain pattern recognition receptors (PRRs) and signal-transducing Toll-like receptors (TLRs) plays a crucial role (5, 26, 40). TLRs in general respond to different types of microbial structures endowing the innate immune response with a relative specificity (36). For example, TLR2 responds to agonists including lipoteichoic acid and lipoproteins, TLR4 responds to enterobacterial lipopolysaccharide (LPS), TLR5 responds to flagellin, and TLR9 responds to bacterial CpG DNA (36).
We have recently investigated interactions between the LT-II toxins and human monocytic THP-1 cells (14). Although LT-IIa and LT-IIb are essentially inactive in stimulating cytokine release, they are both potent regulators of cytokine induction in activated THP-1 cells. Specifically, LT-IIa and LT-IIb downregulate tumor necrosis factor alpha (TNF-α) and interleukin-8 (IL-8) and upregulate IL-1β and IL-10 in THP-1 cells stimulated by bacterial LPS or fimbriae (14). These effects are dependent on their A subunit, since their respective LT-II B pentamers lack regulatory activity. However, both LT-IIaB and especially LT-IIbB induce significant levels of cytokine release in THP-1 cells, in sharp contrast to their respective holotoxins (14).
The objective of this study was to elucidate the mechanisms involved in LT-II B pentamer-induced cellular activation. Although LT-II B pentamers activate nuclear factor (NF)-κB (14), the receptor(s) or signaling pathways involved are uncertain. Gangliosides are constitutively present in lipid rafts, i.e., membrane microdomains involved in the formation and function of signaling receptor complexes, such as hetero-oligomeric PRR-TLR complexes (34, 39). We postulated that the ganglioside-binding property of B pentamers may facilitate possible interactions with PRRs, such as TLRs that are recruited in lipid rafts upon activation with appropriate ligands (26, 39). We have thus investigated whether TLRs participate in LT-II B pentamer-induced cellular activation. We found that the ability of LT-IIaB or LT-IIbB to induce cytokine release in human THP-1 cells or mouse macrophages is significantly dependent upon TLR2 and that their capacity to activate NF-κB in human embryonic kidney (HEK) 293 cells is dependent upon TLR2 cotransfected with TLR1 rather than TLR6. The LT-II B pentamers, however, displayed differential ganglioside-binding requirements for TLR2-mediated cell activation, since only LT-IIbB appeared to require binding to its high-affinity GD1a receptor for this function.
The construction of recombinant plasmids encoding His-tagged versions of wild-type LT-II toxins and their B subunits by means of PCR amplification of appropriate DNA fragments, using oligonucleotide primers with appended His codons (on the reverse primers), has been described previously (14). Similar strategies were used to engineer His-tagged versions of ganglioside-nonbinding mutants of LT-IIaB (with a Thr-to-Ile substitution at position 34) and of LT-IIbB (with a Thr-to-Ile substitution at position 13), using pTDC400/T34I (3) and pTDC700/T13I (4), respectively, as the starting materials. The resulting plasmids, encoding LT-IIaB/T34I and LT-IIbB/T13I, were denoted pHN22 and pHN19, respectively. LPS-free CT was purchased from List Biological Laboratories, Campbell, Calif. The engineering of recombinant B subunit of CT (CTB) with a C-terminal His tag has been described (14).
Recombinant holotoxin or B subunits were expressed in E. coli DH5αF′Kan (Life Technologies, Gaithersburg, Md.) transformed with the appropriate plasmid, and the proteins were extracted from the periplasmic space with polymyxin B treatment (21). The proteins were purified by means of ammonium sulfate precipitation, followed by affinity chromatography using a His·Bind resin column (Novagen, Madison, Wis.) and by size exclusion chromatography using a Sephacryl-100 column and an ΔKTA-FPLC system (Pharmacia, Piskataway, N.J.). Purity was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Fig. (Fig.1)1) and by quantitative Limulus amebocyte lysate assays (using kits from BioWhittaker, Walkersville, Md., or Charles River Endosafe, Charleston, S.C.) to determine endotoxic activity. All holotoxin and B-pentamer preparations were essentially free of incidental LPS contamination (≤0.0064 ng/μg of protein). This was subsequently verified (see Results) in cytokine induction assays, the results of which were unaffected by the presence of the LPS inhibitor, polymyxin B (10 μg/ml). Further evidence against contamination with LPS or other heat-stable contaminants was obtained upon toxin boiling, which destroyed their biological activity (14; also present results).
Human monocytic THP-1 cells (ATCC TIB-202) were differentiated with 10 ng of phorbol myristate acetate/ml for 3 days in 96-well polystyrene culture plates at 37°C in a humidified atmosphere containing 5% CO2. The culture medium consisted of RPMI 1640 (Life Technologies) supplemented with 10% heat-inactivated fetal bovine serum (Life Technologies), 2 mM l-glutamine, 10 mM HEPES, 100 U of penicillin G/ml, 100 μg of streptomycin/ml, and 0.05 mM 2-mercaptoethanol (complete RPMI medium). Differentiated THP-1 cells (1.5 × 105 cells/well) were subsequently washed three times and used in cytokine induction assays. On the basis of earlier experiments (14), pentameric B subunits of LT-II or CT were used at a concentration of 2 μg/ml unless otherwise stated. Stimulation was performed in the absence or presence of blocking monoclonal antibodies (MAbs) to TLR2 (TL2.1), TLR4 (HTA125), or immunoglobulin (Ig) isotype-matched (IgG2a) control (e-Bioscience, San Diego, Calif.). None of the molecules was found to affect cell viability as determined by trypan blue exclusion. Culture supernatants were collected after a 16-h incubation and stored at −80°C until assayed for cytokine content using enzyme-linked immunosorbent assay (ELISA) kits (from eBioscience or Cell Sciences, Canton, Mass.). Similar cell culture procedures were followed to assess cytokine induction (using eBioscience ELISA kits) in mouse peritoneal macrophages from C57BL/6 wild-type mice or mice deficient in TLR2 (37) or TLR4 (17) that have been ninefold backcrossed on the C57BL/6 genetic background. To elicit peritoneal macrophages, mice were injected with 3 to 4 ml of sterile 3% thioglycolate and cells were harvested after 5 days by flushing the peritoneal cavity with 10 ml of ice-cold phosphate-buffered saline four times. Isolated cells were then subjected to density gradient centrifugation (Histopaque 1.083) to remove dead cells and red blood cell contamination. Cells were then washed three times with phosphate-buffered saline and resuspended in complete RPMI medium at a concentration of 106/ml.
HEK 293 cells were plated in 24-well tissue culture plates (5 × 104 cells per well) in 0.5 ml of complete RPMI (as described above except that 2-mercaptoethanol was not included). The cells were incubated for 16 to 20 h after plating at 37°C in 5% CO2 to about 50% confluency. Each well was transfected with 25 ng of pRLnull Renilla luciferase reporter (Promega, Madison Wis.), 75 ng of NF-κB firefly luciferase reporter, and one of the following: empty FLAG-CMV vector alone (100 ng), TLR2 (10 ng) and TLR1 (90 ng), or TLR2 (10 ng) and TLR 6 (90 ng). All the TLRs are N-terminally FLAG tagged derivatives of the human receptors. The DNA mixture was mixed with 5 μl of CaCl2 (2.5 M) and sterile water to a volume of 50 μl, after which 50 μl of 2× HEPES-buffered saline was added. The DNA precipitate was then added dropwise to the cells and incubated for 6 h at 37°C in 5% CO2, after which the media were replaced. Two days after transfection, the cells were stimulated with either no agonist, 20 ng of Pam3Cys-Ser-Lys4 lipopeptide (Pam3Cys; EMC Microcollections, Tuebingen, Germany)/ml, or 2 μg of holotoxin or B pentamer preparations/ml. After 16 h of stimulation, the media were aspirated and 50 μl of passive lysis buffer (Promega) was added to the plates, which were incubated with rocking for 15 min at room temperature. Lysates were transferred to a 96-well plate, and 10 μl of each lysate was evaluated for luciferase activity using a dual-luciferase reporter assay system. (Promega). Each firefly luciferase value was divided by the Renilla value to correct for transfection efficiency. All corrected values were normalized to that of the no-agonist control, whose value was taken as 1.
Activation of the transactivating p65 subunit of NF-κB in THP-1 cells was determined by means of an NF-κB/p65 transcription factor assay kit (Active Motif, Carlsbad, Calif.) (11, 12). This is an ELISA-based procedure in which the detecting antibody recognizes an epitope of NF-κB p65 that is accessible only when NF-κB is activated and bound to its target DNA (containing the NF-κB consensus binding site 5′-GGGACTTTCC-3′) attached to 96-well plates. Extract preparation and NF-κB/p65 ELISA were carried out according to protocols supplied by the manufacturer. The optimal time of stimulation (90 min) and amount of total protein (7.5 μg) used in the ELISA were determined in earlier experiments (14).
Data were evaluated by analysis of variance and the Dunnett multiple-comparison test (see Fig. Fig.22 and and44 and Table Table1),1), using the InStat program (GraphPad Software, San Diego, Calif.). Statistical differences were considered significant at the level of P values of <0.05. These experiments were performed in triplicate and were repeated twice for verification. A nonparametric procedure was used to analyze the data from the luciferase gene reporter assays (see Fig. Fig.3)3) because of significant differences among the standard deviations of the groups under comparison. Specifically, the data from four independent but similar assays were pooled and analyzed by a professional biostatistician, using the multiresponse permutation procedure for randomized block experiments. The analysis was performed using a FORTRAN program distributed by the developer of the methodology, Paul W. Mielke, Jr., at Colorado State University (23). All experimental groups were compared with the no-agonist control for TLR1/TLR2 or TLR2/TLR6 activation. The analysis also included a comparison of TLR1/TLR2 versus TLR2/TLR6 activation by the same agonists. Hypothesis testing was performed at the 0.05 significance level.
Several microbial proteins appear to display molecular patterns that can activate cells through TLR2 or TLR4 (9, 11, 13, 18, 19, 22, 35). Whether LT-II B-pentamer-induced cellular activation is similarly dependent on TLRs was initially addressed in cytokine induction assays with THP-1 cells and anti-TLR MAbs. We found that induction of IL-8 by LT-IIaB, LT-IIbB, or CTB was partially but significantly (P < 0.05) inhibited by a MAb to TLR2 (Fig. (Fig.2A).2A). CTB was also used at a twofold-higher concentration (4 μg/ml) to enhance induction of IL-8 and thereby to improve evaluation of the inhibitory effect (Fig. (Fig.2A,2A, insert). Anti-TLR4 MAb or an isotype control had no significant effect on IL-8 induction by the B pentamers (Fig. (Fig.2A2A and insert). Similarly, IL-1β induction by LT-IIaB or LT-IIbB was significantly (P < 0.05) inhibited by anti-TLR2 but not by anti-TLR4 or the isotype control (Fig. (Fig.2B)2B) (CTB was not tested, since it does not induce measurable IL-1β (14). Likewise, anti-TLR2 but not anti-TLR4 inhibited induction of TNF-α and IL-6 release by LT-IIbB (Fig. (Fig.2C);2C); LT-IIaB and CTB were not tested because they do not induce significant release of these cytokines (14). The inhibitory effect of anti-TLR2 MAb was also significant (P < 0.05) in comparison to treatment with anti-TLR4 MAb in the case of LT-IIaB (Fig. 2A and B) or LT-IIbB (Fig. 2A to C). However, in the case of CTB, the TLR2 MAb effect was not significantly different from that of anti-TLR4 (Fig. (Fig.2A2A and insert). We have thus sought additional, independent approaches to conclusively confirm the role of TLRs in B-pentamer-induced cellular activation (see below), as suggested by the TLR MAb data. The degree of effectiveness of the blocking anti-TLR MAbs was monitored in cytokine induction assays using established TLR2 (Pam3Cys) and TLR4 (Ec-LPS) agonists; the obtained results confirmed the specificity of the MAbs, although their inhibitory effect was not complete (Fig. (Fig.2D2D).
Induction of IL-8 release in THP-1 cells by 2 μg of LT-IIaB (IL-8 response, 4,752 ± 611 pg/ml), LT-IIbB (IL-8 response, 28,530 ± 4,367 pg/ml), or CTB (IL-8 response, 704 ± 84 pg/ml)/ml was unaffected in the presence of 10 μg of polymyxin B/ml (corresponding IL-8 responses: 4,459 ± 489 pg/ml, 30,530 ± 3,005 pg/ml, and 789 ± 92 pg/ml, respectively) but was abolished upon boiling of the B pentamers (corresponding IL-8 responses: 147 ± 45 pg/ml, 132 ± 64 pg/ml, and 108 ± 28 pg/ml, respectively). Conversely, when THP-1 cells were activated by 0.2 μg of E. coli LPS/ml, the induced IL-8 release (34,839 ± 3187 pg/ml) was inhibited by polymyxin B (3,098 ± 618 pg/ml) but not by boiling the LPS (37,122 ± 5,890 pg/ml). These findings verify that activation of cells by B pentamers was not attributable to contamination with LPS or other heat-stable contaminants.
To obtain further support for TLR2 involvement in B-pentamer-induced cellular activation, we used HEK 293 cells transiently cotransfected with cDNAs encoding TLR2 with either TLR1 or TLR6, both of which have been shown to cooperate with TLR2 to mediate signaling (25). Specifically, HEK 293 cells transfected with TLRs or “empty” control vector were stimulated with LT-IIaB, LT-IIbB, CTB, or their respective holotoxins. Pam3Cys, a synthetic TLR2 agonist (15), was used as a positive control. All cotransfections included a cDNA encoding firefly luciferase driven by an NF-κB-dependent promoter in order to monitor cellular activation. We found that besides Pam3Cys, only LT-IIaB and LT-IIbB induced significant (P < 0.05) cellular activation upon transfection with TLRs (Fig. (Fig.3).3). LT-IIaB activated only TLR1/TLR2-transfected cells (Fig. (Fig.3).3). LT-IIbB additionally activated TLR2/TLR6-transfected cells, although it displayed a significantly higher (P < 0.05) capacity to activate cells cotransfected with TLR1 plus TLR2 (Fig. (Fig.3).3). The ability of LT-IIaB or LT-IIbB to activate HEK 293 cells was diminished when these were transfected with TLR2 alone (not shown). None of the holotoxins induced significant TLR-dependent activation in HEK 293 cells (Fig. (Fig.3),3), in line with their weak cytokine-inducing capacity observed in earlier experiments using THP-1 cells (14). As expected, a TLR4 agonist (E. coli LPS) did not activate either TLR1/TLR2- or TLR2/TLR6-transfected cells (data not shown). These results demonstrate a TLR2 requirement in cellular activation by LT-IIaB or LT-IIbB and suggest that TLR1 may be a signaling partner of TLR2 in this regard. This is the first demonstration that enterotoxin B pentamers cause cellular activation in a TLR-dependent fashion.
If TLR2 plays a major role in mediating cytokine release by LT-IIaB or LT-IIbB, then its absence in mutant mouse macrophages would be expected to result in significant reduction of this activity of LT-II B pentamers. We have thus evaluated the ability of LT-IIaB or LT-IIbB to induce cytokine release in TLR2-deficient macrophages compared with wild-type or TLR4-deficient cells. Known TLR agonists (Pam3Cys [TLR2] and E. coli LPS [TLR4]) were used as positive or negative controls. All control TLR agonists and LT-II B pentamers induced release of TNF-α (Fig. (Fig.4A)4A) or IL-6 (Fig. (Fig.4B)4B) in wild-type macrophages. In results similar to those with Pam3Cys, however, neither LT-IIaB nor LT-IIbB could stimulate substantial cytokine release in TLR2-deficient macrophages, although they were unaffected by TLR4 deficiency (Fig. (Fig.4).4). As expected, the reverse was true for E. coli LPS, which maintained its cytokine-inducing ability in TLR2-deficient macrophages but not in TLR4-deficient macrophages (Fig. (Fig.4).4). These results demonstrate that TLR2 is required for LTIIaB- or LTIIbB-induced activation of mouse macrophages and reinforce similar findings obtained with human cell lines (Fig. (Fig.22 and and33).
Since TLRs often require cooperation with other PRRs to mediate cellular activation, we determined whether ganglioside binding is required for the ability of LT-IIaB or LT-IIbB to induce TLR2-dependent activation of THP-1 cells. For this purpose we used two mutants, LT-IIaB/T34I and LT-IIbB/T13I, which show no binding to any gangliosides tested, such as GD1a, GD1b, GT1b, GQ1b, GM1, GM2, or GM3 (3, 4). Surprisingly, we found that LT-IIaB/T34I was even more effective than the wild-type molecule in inducing cytokine release or NF-κB p65 activation (Table (Table1).1). Therefore, whereas TLR2 is important for cellular activation by LT-IIaB (Table (Table1),1), gangliosides (at least the ones mentioned above, including those important for LT-IIa toxicity) do not play a role in this regard. On the other hand, the LT-IIbB/T13I mutant did not retain any of the proinflammatory activity (cytokine induction or NF-κB p65 activation) (Table (Table1)1) of the wild-type molecule. Therefore the high-affinity ganglioside receptor of LT-IIbB, GD1a, also appears to be required for the ability of this molecule to activate THP-1 cells in a TLR2-dependent mode (Table (Table1).1). Alternatively, the tyrosine at position 13 of LT-IIbB, which is critical for GD1a binding, may also be required for stimulation of TLR2.
The results of this study strongly implicate TLR2 in cellular activation by the B pentamers of LT-II enterotoxins. In contrast, the LT-IIa and LT-IIb holotoxins do not significantly activate TLR1/TLR2- or TLR2/TLR6-transfected HEK 293 cells (Fig. (Fig.3),3), nor do they induce significant cytokine release in THP-1 cells (14). It appears, therefore, that the ability of LT-II B pentamers to induce TLR2-dependent NF-κB activation and cytokine release is facilitated by the absence of the A subunit. In this respect, the capacity of the LT-IIb B pentamer to activate NF-κB and induce proinflammatory cytokine release is strongly antagonized by the LT-IIb holotoxin (14). These biological differences between the LT-II holotoxins and their B pentamers may help elucidate distinct immunomodulatory mechanisms, which may be at least as complex as those of type I holotoxins and their respective B pentamers (28). An adjuvant mechanism whereby the LT-II B pentamers may promote adaptive immune responses is possibly through induction of TLR2-dependent inflammatory and immunoregulatory activity. In fact, TLRs are considered adjuvant receptors that initiate adaptive immunity through induction of cytokines and upregulation of the B7 costimulatory molecules (36). In this regard, we have previously shown that LT-IIaB upregulates expression of B7-1 but not of B7-2 on mouse antigen-presenting B cells, while the reverse is true for CTB (LT-IIbB was not examined) (20). Since activated B cells express TLR2 (7), it becomes important to determine whether LT-IIaB upregulates costimulatory molecules in these or other antigen-presenting cell types in a TLR2-dependent way. In contrast to the LT-II B pentamers, the adjuvant activity of the LT-II holotoxins (21) may be exerted independently of TLR activation and under relatively noninflammatory conditions, as suggested for CT in an in vivo model (38).
Since anti-TLR2 displayed a partial inhibitory effect on LT-II B pentamer-induced cytokine release in THP-1 cells, it was initially concluded that there may be an additional signaling mechanism involved. However, the essentially complete unresponsiveness of TLR2-deficient mouse macrophages to LT-II B pentamers suggests that either the anti-human TLR2 MAb used was not completely efficient or that there may be differences between the human and mouse models used. The LT-II B pentamers appeared to be relatively more proinflammatory in the mouse model than in the human model, as suggested by their comparison with established TLR agonists in the present mouse cell experiments (Fig. (Fig.4)4) and earlier human THP-1 cell experiments (14). Moreover, LT-IIbB was consistently more proinflammatory than LT-IIaB in the human model (Fig. (Fig.22 and and3;3; Table Table1),1), whereas in the mouse model LT-IIaB was stronger in TNF-α induction and LT-IIbB was stronger in IL-6 induction (Fig. (Fig.44).
Gangliosides are considered to be prominent receptors for type I and II enterotoxins (16). Our findings support the notion that ganglioside receptors are not exclusively responsible for the immunomodulatory activities of LT-II B pentamers. It is of interest also that certain mutant B pentamers of CT (H57A) or LT-I (H57S), which retain high-affinity binding to GM1, are nevertheless defective in immunomodulatory signaling in lymphocytes (1, 8). It could be surmised that the structural alterations in these mutant molecules, while not preventing binding to GM1, may preclude interactions with additional receptors required for signaling. Apparently, ganglioside binding alone is not sufficient to mediate the immunomodulatory effects of type I enterotoxin B pentamers. These observations are consistent with the recently developed concept that cellular activation by microbial molecules involves interactions with several cooperating host receptors, whereas microbial interactions with a single receptor often represent an oversimplified model. Specifically, innate immune recognition and signaling involve dynamic and transient formation of receptor complexes within lipid rafts (26, 39, 40), which, if disrupted by certain drugs like nystatin, are unable to support cytokine induction (39). At least seven different PRRs have been implicated in participating in LPS signaling receptor complexes (5, 26, 32, 40). Moreover, Porphyromonas gingivalis fimbriae induce cellular activation through interactions with at least three PRRs (11, 13). Similarly, multiple receptors may also be implicated in innate immune cell activation by LT-IIbB. This B pentamer binds with high affinity to GD1a (10), which may be involved in NF-κB activation and cytokine induction, since these activities are not retained by the GD1a-nonbinding mutant (T13I) of LT-IIbB (Table (Table1).1). The same LT-IIbB activities are also dependent on TLR2 (Table (Table11 and Fig. Fig.22 to to4).4). Therefore, it appears that both GD1a and TLR2 are important in cellular activation by LT-IIbB. Moreover, our findings suggest that the signaling partner of TLR2 may be TLR1, since LT-IIbB activated TLR1/TLR2-transfected HEK 293 cells whereas substituting TLR6 for TLR1 resulted in reduced activation (Fig. (Fig.33).
A glycosphingolipid that binds E. coli P fimbriae through a Galα1-4Galβ disaccharide was previously shown to facilitate activation of TLR4 by P fimbriae (9). More recently, and while these studies were in progress, it was reported that certain gangliosides (GD1a, GD1b, and GT1b among eight tested) serve as coreceptors with TLR5 for induction of human β-defensin-2 by Salmonella enterica serovar Enteritidis flagellin in Caco-2 cells (24). It is thus becoming apparent that gangliosides and glycosphingolipids in general may have the potential to cooperate with TLRs for activating signaling pathways. We speculate that GD1a may somehow place LT-IIbB within proximity of interaction with TLR2 in lipid rafts, in a mode analogous to the way in which CD14 facilitates LPS recognition by TLR4. Alternatively, the propensity of TLR2 to recognize and respond to certain lipids attached to proteins or peptides (36) may also enable it to also recognize GD1a-bound LT-IIbB as a lipid-protein complex.
In contrast to LT-IIbB, the LT-IIaB variant did not appear to depend on its high-affinity-binding gangliosides for THP-1 activation (Table (Table1).1). Rather, we observed enhanced cytokine induction by the ganglioside-nonbinding LT-IIaB/T34I mutant relative to that by the wild-type molecule. This may be due to LT-IIaB/T34I being free to interact with receptors that promote cytokine induction, as opposed to the wild-type molecule, which may preferentially bind to its high-affinity GD1b receptor or other gangliosides. An alternative, though not mutually exclusive, explanation might be that GD1b mediates negative regulatory signals for cytokine induction that are absent in LT-IIaB/T13I-activated cells. In brief, although binding to GD1b is essential for the toxic activity of the LT-IIa holotoxin (3), it is not required for cytokine induction by the LT-IIa B pentamer. LT-IIaB may instead use other PRRs for that matter or interact directly with the TLR2 complex. CTB was by far the weakest among the B pentamers tested in cytokine induction, and it is not surprising that we were unable to detect significant TLR2 activation in HEK 293 cells (Fig. (Fig.3).3). Whether its ganglioside receptor, GM1, can function as a TLR coreceptor is unknown, although this ganglioside was recently shown to serve as a functional coreceptor for fibroblast growth factor 2 (31).
Although it is widely accepted that TLRs have evolved to detect and respond to pathogen-associated molecular patterns (36), there are examples suggesting that certain pathogens have evolved virulence mechanisms that exploit TLRs. For instance, the Yersinia virulence antigen, LcrV, induces TLR2-dependent immunosuppression through IL-10 release (35). Moreover, certain TLR2 agonists induce a Th2-biased immune response, which may compromise pathogen clearance due to diminished production of biologically active IL-12 (29). Nevertheless, possible TLR2-induced Th2 polarization may rather protect against enteropathogens through effective induction of intestinal IgA antibodies that neutralize heat-labile enterotoxins (28, 33). Furthermore, it is noteworthy that compared to proinflammatory cytokines, LTII B pentamers induce very little IL-10, and although LT-IIb holotoxin synergizes with LT-IIb pentamer for IL-10 release, this is not a TLR2-mediated effect (14). Thus, at present there is no reason to suggest that LT-II B pentamers have the capacity to subvert TLR activation for the enhancement of pathogen virulence.
These findings introduce TLRs as a new level of complexity in host recognition of heat-labile enterotoxins. The fact that all toxin molecules tested here were expressed, isolated, and analyzed for purity in the same exact way, yet they displayed dramatic differences in cellular activation capabilities, shows that TLR2 activation and cytokine induction cannot be attributed to unidentified contaminants in our preparations. In this connection, the most striking finding was the total cellular activation inability of the LT-IIbB/T13I point mutant in comparison to the wild-type molecule. The B subunits of type II toxins are thus genuine TLR2 agonists. Hitherto, the adjuvant effect of heat-labile toxins has been thought to depend on their ganglioside-binding and toxic-enzyme activities, though the extent of the latter is debatable in view of various results obtained with inactive mutants of type I toxins (28, 33). In contrast, the B subunits of type II toxins are immunostimulatory by mechanisms that depend on TLRs but not necessarily on ganglioside recognition. As a result, we suggest that type II enterotoxin B subunits have useful properties as adjuvants that may avoid some of the complications surrounding the use of type I toxins or their B subunits.
This work was supported by U.S. Public Health Service grants DE015254, AI052344, DE06746, and DE13833 from the National Institutes of Health.
We thank Donald Mercante (LSU Health Sciences Center, Center of Excellence in Oral and Craniofacial Biology) for expert statistical analysis.
Editor: J. T. Barbieri