ABSTRACT
The diarrheal mechanisms in Aeromonas enteritis are not completely understood. In this study we investigated the effect of aeromonads and of their secretory products on ion secretion and barrier function of monolayers of human intestinal cells (HT-29/B6). Ion secretion was determined as a short-circuit current (ISC) of HT-29/B6 monolayers mounted in Ussing-type chambers. Transepithelial resistance (Rt) served as a measure of permeability. A diarrheal strain of Aeromonas hydrophila (strain Sb) added to the mucosal side of HT-29/B6 monolayers induced a significant ISC (39 ± 3 μA/cm2) and decreased the Rt to ∼10% of the initial value. A qualitatively identical response was obtained with sterile supernatant of strain Sb, and Aeromonas supernatant also induced a significant ISC in totally stripped human colon. Tracer flux and ion replacement studies revealed the ISC to be mainly accounted for by electrogenic Cl− secretion. Supernatant applied serosally completely abolished basal ISC. The supernatant-induced ISC was inhibited by the protein kinase C inhibitor chelerythrine, whereas a protein kinase A inhibitor (H8) and a Ca2+ chelator (BAPTA-AM) had no effect. Physicochemical properties indicated that the supernatant's active compound was an aerolysin-related Aeromonas beta-hemolysin. Accordingly, identical ISC and Rt responses were obtained with Escherichia coli lysates harboring the cloned beta-hemolysin gene from strain SB or the aerA gene encoding for aerolysin. Sequence comparison revealed a 64% homology between aerolysin and the beta-hemolysin cloned from Aeromonas sp. strain Sb. In conclusion, beta-hemolysin secreted by pathogenic aeromonads induces active Cl− secretion in the intestinal epithelium, possibly by channel insertion into the apical membrane and by activation of protein kinase C.
Numerous reports have implicated Aeromonas species as a cause of human enteric disease in children as well as in adults (for reviews, see references 3, 22, 23, 28, and 36). However, although a large body of literature on Aeromonas virulence factors has been accumulated (for a review, see references 9 and 22), the mechanisms of the diarrheic action of Aeromonas are not completely understood. As for other pathogens, Aeromonas virulence factors can be divided into cellular properties such as adherence, invasiveness, motility, and extracellular factors secreted by pathogenic aeromonads (22). Evidence for the enteropathogenic potential of these virulence factors was obtained from their epidemiological association with diarrheal disease (3, 10, 24), from their ability to induce enteral fluid secretion in animal models, and from the observation of effects linked to enteral fluid secretion elicited in various animal and cell culture systems (see, for example, references 4, 7, 29, and 33). Surprisingly, then, the precise mechanism for how Aeromonas infection causes intestinal fluid secretion has not yet been elucidated.
From a general pathophysiological point of view, intestinal pathogens can induce diarrhea by induction of active ion secretion and/or by impairment of the epithelial barrier function of the intestine. In support of the latter mechanism, certain Aeromonas toxins were shown to be cytotoxic for intestinal cells, as well as for other mammalian cells (1, 6, 19, 29). However, it is questionable whether massive physical disruption of the epithelial barrier as the result of necrosis of enterocytes is the predominant mode of action in Aeromonas enteritis, which usually presents as watery diarrhea but only rarely as dysenteric disease (3). Therefore, it seemed to be necessary to take a closer look at the interaction of the bacteria and their toxins with the intestinal epithelium in order to define the mechanisms of diarrhea caused by aeromonads. However, to the best of our knowledge, at present there is only one study addressing the issue of intestinal barrier impairment by an Aeromonas toxin at the enterocyte level (2). In that study, Aeromonas beta-hemolysin caused a decrease in transepithelial resistance in polarized intestinal epithelial cells, but no attempt was undertaken to characterize the mechanism underlying this resistance drop, and no data referring to transepithelial movement of ions or other solutes were provided. Therefore, it is currently not known whether Aeromonas induced diarrhea is due solely to passive solute flux after damage of the epithelial barrier or also to active ion secretion. Furthermore, the effector mechanism of the Aeromonas induced barrier impairment remains to be identified.
In order to gain more insight into the epithelial effects of intestinal Aeromonas infection, we performed electrophysiological studies in monolayers of HT-29/B6 cells. These cells represent a stable, highly differentiated subclone derived from wild-type HT-29 cells by glucose deprivation (27). When grown on permeable supports, they form polarized monolayers with intact regulation of tight junction formation, which is the key determinant of epithelial barrier function (38). In addition, HT-29/B6 monolayers are capable of regulated vectorial Cl− and mucin secretion (12, 13). Thus, this highly differentiated cell line possesses most of the features of native intestinal epithelia necessary for maintenance of intestinal barrier function, as well as for the regulation of intestinal ion transport. For these reasons, HT-29/B6 cells have been used as model epithelium for mechanistic investigations of altered intestinal ion secretion and barrier function in many studies before (see, for example, references 12, 13, 31, and 38).
In the present study, we investigated the ability of aeromonads to stimulate ion secretion and to impair epithelial barrier function of HT-29/B6 monolayers. In an initial set of experiments, an (apparently) enteropathogenic Aeromonas strain, isolated as the only enteropathogen from a patient with acute watery diarrhea, was compared to an (apparently) nonenteropathogenic strain that was isolated from normal stool of a patient without any symptoms of diarrhea. Since only the enteropathogenic strain altered the transport and barrier properties of the monolayers, further experiments focused on the mechanisms and signal transduction of the effects elicited by the pathogenic strain and its secretory products.
MATERIALS AND METHODS
Bacterial strains and plasmids.All Aeromonas strains were obtained from the Department of Microbiology of the Freie Universität Berlin. Aeromonas sp. strain Sb was isolated from a patient with acute watery diarrhea. Other than Aeromonas, no other enteropathogens were found in the stool specimens of this patient. Strain Ha has been isolated from the normal stool of a patient without diarrhea. Sequencing of the 16S rRNA subunit identified strain Sb as genotype Aeromonas hydrophila and strain Ha as genotype Aeromonas veronii. Escherichia coli DH5α and TOP10F′ (Invitrogen, Karlsruhe, Germany) were used for plasmid expression. Plasmid pBR322 encoding for ampicillin and tetracycline resistance and harboring the aerA gene of a diarrheal Aeromonas sobria isolate (21) was provided by T. Chakraborty. Plasmid pCR2.1 TOPO (Invitrogen) was used for cloning of the beta-hemolysin gene of strain Sb.
Cell culture.HT-29/B6 cells were cultured as described previously (12, 27). For electrophysiological experiments, cells were seeded on polycarbonate filters fixed at the bottom of culture plate inserts (Millipore-PCF, effective membrane area of 0.6 cm2; Millipore, Eschborn, Germany). Measurement of the transepithelial resistance (Rt) was used to test for confluence. After confluence, monolayers of HT-29/B6 cells form an epithelial barrier between the apical compartment within the insert and the basolateral compartment outside. Monolayers were used for the experiments 7 or 8 days after seeding, when Rt values were between 300 and 500 Ω cm2.
Measurements in human colon.Segments of macroscopically normal colon were obtained from five patients undergoing resective surgery for colon carcinoma. After removal of underlying subepithelial layers (“total strip”) as described earlier (11), the colon epithelium was mounted into Ussing-type chambers. Then, electrophysiological studies were performed as described below for the cell culture experiments.
Electrophysiological studies.Intact culture plate inserts carrying confluent filter grown monolayers of HT-29/B6 cells were placed into modified Ussing-type chambers as described previously (12). Briefly, the mucosal and the serosal compartment were filled with 10 ml of a modified Ringer solution (bathing solution) containing 151 mM Na+, 5 mM K+, 1.7 mM Ca2+, 0.9 mM Mg2+, 130.4 mM Cl−, 28 mM HCO3−, 1 mM H2PO4−, 0.9 mM SO42−, and 25 mM d-(+)-glucose. The solution was stirred and oxygenated via bubble lift. The pH at 95% O2 and 5% CO2 was 7.4. The temperature was kept constant at 37°C by means of temperature-controlled water jackets. For experiments on the direct interaction between aeromonads and HT-29/B6 cells, single colonies of the respective Aeromonas isolate were picked from blood agar plates, dissolved in RPMI 1640 and added to the mucosal compartment (1 colony per 10 ml of bathing solution, bacterial concentration ca. 105 CFU/ml as determined by plating after serial dilution). For investigating the effects of Aeromonas secretory products, single colonies of the respective strains were picked from blood agar plates and suspended in RPMI 1640 (one colony per milliliter). The bacterial suspension was diluted with bathing solution to 10% (vol/vol), incubated at 37°C, and gassed with 95% CO2-5% O2. After 12 h of incubation the bacteria were removed by centrifugation, followed by sterile filtration. The experiment was then started by addition of the filtrate to the bathing solution of HT-29/B6 monolayers placed in Ussing chambers. If not otherwise stated, the filtrate was added to the mucosal compartment, imitating the luminal access of bacterial products during enteric Aeromonas infection. For controls, an equal amount of carrier (bathing solution with 10% RPMI 1640) was added to the bathing solution. In other experiments, sterile lysates of DH5α or TOP10F′ cells were added to the mucosal compartment as indicated. Short-circuit current (ISC) and transepithelial resistance (Rt) were determined by a computerized automatic clamp device (Fiebig Hard- and Software, Berlin, Germany). Mucosal-to-serosal (ms) mannitol flux studies were performed with [3H]mannitol as described previously (39). Ion transport was determined by measurements of unidirectional (i.e., ms and serosal-to-mucosal [ms and sm, respectively]) 22Na and 36Cl fluxes (DuPont, Bad Homburg, Germany). For calculating net fluxes, monolayers were matched for conductance. All flux experiments were performed under short-circuit conditions. For ion replacement studies, the monolayers were stimulated with Aeromonas supernatant as described above. When submaximum ISC values were reached, 5 ml of the bathing solution was cautiously removed and replaced by a modified bathing solution free of Cl− ions (see below). In this manner, the bubble lift-driven circulation of the bathing solution was uninterrupted. This procedure was repeated three times. By then, the Cl− content of the bathing solution was reduced to 16.3 mM. ISC values were determined after a 10-min equilibration period. Then, a high Cl− concentration within the bathing solution was restored by repeated (three times) partial exchange of the bathing solution with standard solution again. The Cl−-free solution was composed of 60 mM Na2SO4, 2.7 mM K2SO4, 1.2 mM MgSO4 · 7H2O, 1.2 mM CaSO4 · 2H2O, mM 2.4 Na2HPO4, 0.6 mM NaH2PO4 · H2O, 10 mM Tris, 10 mM HEPES, and 25 mM d-(+)-glucose. The pH of the solution was adjusted to 7.4 by titration with H2SO4. For all electrophysiological experiments, independent controls were processed in parallel.
Immunofluorescence.In order to test for the integrity of the tight junctional meshwork, the tight junction protein ZO-1 was visualized after 2 h of Aeromonas exposure, when Rt had leveled off at minimum values. Monolayers were removed from the Ussing chamber washed with phosphate-buffered saline (PBS) and fixed in ice-cold methanol for 10 min. The cells were then washed with PBS and permeabilized by incubation with 0.5% Triton X-100 for 5 min. To reduce unspecific binding sites, cells were incubated with 0.5% goat serum for 30 min at room temperature. The cells were then incubated with mouse anti-ZO-1 antibodies (1:50; Life Sciences Research, Heidelberg, Germany), rabbit anti-occludin antibodies, or anti-claudin-1 antibodies (Zymed, San Francisco, Calif.) for 30 min at room temperature, followed by a wash with 0.5% goat serum. Alexa Fluor immunofluorescence antibodies (Molecular Probes Europe, Leiden, The Netherlands) Alexa Fluor-594 goat anti-rabbit immunoglobulin G and Alexa Fluor-488 goat anti-mouse immunoglobulin G (1:500) were added for 30 min. After the samples were washed and mounted in ProTaqs MountFluor medium (Biocyc, Luckenwalde, Germany), fluorescence microscopy was performed.
LDH release assay.Lactate dehydrogenase (LDH) release from HT-29/B6 cells was measured according to the method of Madara and Stafford (30). Briefly, the LDH content in the supernatant of controls and of cells treated with Aeromonas supernatant was determined and compared to the total LDH content of the residual cells which was determined after detergent extraction with 2% Triton X-100 for 30 min. LDH release was expressed as a percentage of total LDH released into the supernatant.
Measurement of hemolytic activity.To assess the hemolytic activity of bacterial lysates or culture supernatants, a serial twofold dilution of the respective samples (in PBS) was incubated with 0.5% sheep erythrocytes (diluted with PBS containing 10 mM dithiothreitol) in microtiter trays at 37°C. Sodium dodecyl sulfate was used as a positive control, and PBS served as a negative control. A hemolytic reaction was considered positive if lysis of erythrocytes was evident on inspection after 4 h of incubation. The hemolytic activity of a sample was expressed as the reciprocal of the highest dilution that still gave a positive hemolytic reaction.
Cloning of the beta-hemolysin gene of strain Sb.Total DNA was extracted from strain Sb with 0.05 N NaOH at 95°C, followed by neutralization with 1 M Tris-HCl and removal of cell debris by centrifugation. The beta-hemolysin gene was then amplified by PCR. Seven different primer pairs derived from known Aeromonas beta-hemolysin sequences published in GenBank were tested. The best amplification results were obtained with the primers SE-F (5′-GAAGTGATCAATCCGGAAGA-3′) and SE-R (5′-CTATGAAAGGGGCCTGCG-3′), which were based on an aerolysin-related beta-hemolysin of Aeromonas hydrophila strain NLEP A-1607 (GenBank accession number AF410466 ). The PCR product was purified by using a PCR purification kit (Qiagen, Hilden, Germany) and cloned into plasmid pCR2.1 TOPO (Invitrogen) encoding ampicillin and kanamycin resistance by using the TOPO TA cloning kit (Invitrogen) according to the instructions provided. As confirmed by sequencing, the insert was located out of frame with the ATG initiation codon of the β-galactosidase gene. E. coli TOP10F′ cells transformed with the recombinant plasmid were plated on ampicillin-containing X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) plates. Fifty white colonies were picked and plated on kanamycin-containing blood agar plates. A colony showing a distinct zone of beta-hemolysis was cultured overnight in Luria-Bertani medium containing ampicillin. The plasmid DNA was then isolated with a commercially available kit (GFX Micro Plasmid Prep Kit; Amersham Biosciences, Freiburg, Germany), and the insert was analyzed by sequencing.
Lysates from E. coli. E. coli DH5α or TOP10F′ cells as indicated were harvested from overnight cultures (250 ml), washed with PBS, and resuspended in 10 ml of ice-cold bathing solution. The intracellular content of the bacteria was released into the bathing solution after cell wall disrupture with a hydraulic press (Thermo Spectronic, Cambridge, United Kingdom). After cell lysis, cell debris and cell membrane remnants were removed by centrifugation (20,000 × g for 30 min at 4°C), and the lysate was used for electrophysiological experiments and hemolysin titer tests.
RESULTS
Effect of coincubation with Aeromonas on transport and barrier properties of HT-29/B6 monolayers.Bacteria were cultured in RPMI 1640 and added to the mucosal compartment of HT-29/B6 cells mounted in Ussing chambers to yield an initial concentration of 105 CFU/ml. By 90 min after addition of the pathogenic strain Sb (diarrheal isolate), the ISC started to increase and the Rt values started to decline (Fig. 1). Maximum ISC values of 39 ± 3 μA/cm2 were reached after 140 min of coincubation. At this time point Rt had dropped to ca. 50% of its initial value. Thereafter, the ISC gradually dropped to ∼20 μA/cm2. Rt values, on the other hand, decreased continually until they leveled off after 200 min at a residual Rt of ∼24 Ω cm2 equaling 6.8% of the initial Rt (354 ± 4 Ω cm2). In contrast to strain Sb, the apathogenic strain Ha (derived from an asymptomatic patient without gastrointestinal symptoms) did not evoke any effects on ISC and Rt over more than 5 h, although both strains grew well during the experiment. Thus, after 3 h of incubation in the bathing solution, serial dilutions revealed an estimated bacterial number of 4 × 105 CFU/ml for strain Sb compared to 8 × 105 CFU/ml for strain Ha. The respective numbers after 5 h of incubation were 106 CFU/ml for strain Sb and 8 × 106 CFU/ml for strain Ha.
Time course of ISC and Rt after coincubation of HT-29/B6 monolayers with Aeromonas. Pathogenic (strain Sb) or apathogenic (strain Ha) aeromonads were added to the mucosal bathing solution to yield an initial concentration of 105 CFU/ml. Registration was then started. After a 90-min latency period, the pathogenic strain Sb induced a marked increase in ISC (upper panel) and a decrease in Rt (lower panel). In contrast, the apathogenic strain Ha left ISC and Rt unaltered over the whole 5-h registration period. Values are means ± the standard error of the mean (SEM) of six monolayers for each group.
Effect of Aeromonas supernatant on transport and barrier properties of HT-29/B6 monolayers.In order to determine whether the effects observed were due to direct interaction of the bacteria with the epithelial cells or rather caused by a toxin secreted into the bathing medium, monolayers were coincubated with Aeromonas sp. strain Sb as for the experiments described in Fig. 1. After a 100-min lag time, the ISC started to increase. When maximum ISC values were reached (Fig. 2A; ISC = 41 ± 2 μA/cm2 and Rt = 63% ± 4% of initial Rt [Rt data not shown]), the mucosal bathing solution was removed and cleared from bacteria by centrifugation and sterile filtration. By addition of the bacterium-free solution to the mucosal side of new monolayers mounted in Ussing chambers, without any latency, the Aeromonas effect on ISC and Rt could be restored (Fig. 2B, maximum ISC = 29 ± 1 μA/cm2 and Rt after 120 min = 68% ± 9% of initial Rt). Obviously, the active compound within the bathing solution was independent from the presence of bacteria and, as a consequence, the ISC and Rt responses must have been caused by the action of a soluble factor secreted by the aeromonads. Subsequent experiments were therefore performed with bacterium-free supernatant.
Effect of sterile Aeromonas supernatant on ISC and Rt of HT-29/B6 monolayers. (A) Strain Sb aeromonads were added to the mucosal bathing solution to yield an initial concentration of 105 CFU/ml. After a 100-min latency, induction of ISC became evident. When maximum ISC values were reached, the apical bathing solution containing the bacteria (i.e., the supernatant) was removed and cleared from the aeromonads by centrifugation and sterile filtration. The sterile supernatant was then added to the mucosal compartment of new HT-29/B6 monolayers (indicated by the arrow). (B) Immediately after addition of the supernatant, the ISC started to increase and, after a short latency period, the Rt started to decrease again.
In Fig. 3, dose-response data obtained with different concentrations of Aeromonas supernatant are depicted. Both the maximum ISC values and the Rt drop induced clearly correlated to the concentration used. Next, we tested whether the monolayers were also sensitive to serosal addition of Aeromonas sp. strain Sb supernatant. In fact, the epithelial resistance decreased faster with serosal versus mucosal addition of the supernatant (Rt at 50% of initial value, 22 min after serosal addition versus 54 min after mucosal addition, respectively). Interestingly, after serosal addition the basal short circuit current was completely abolished, indicating cessation of all active electrogenic ion transport processes (Fig. 4).
Dose response of sterile Aeromonas supernatant on ISC and Rt of HT-29/B6 monolayers. Supernatant from overnight cultures of strain Sb was added to the mucosal compartment of monolayers mounted in Ussing chambers at dilutions of up to 1:10,000. The magnitude of the maximum ISC induced (A) and the decrease in Rt (B) were clearly dependent on the concentration of the supernatant. Rt was determined for comparison 30 min after addition of the supernatant. Values are means ± the SEM of five to six monolayers for each concentration. ❋❋❋, P < 0.001 versus controls treated with carrier only.
Effect of serosal application of Aeromonas supernatant on ISC of HT-29/B6 monolayers. Supernatant (dilution 1:10) from an overnight culture of strain Sb was added to the mucosal (dotted line) or serosal (straight line) compartment of monolayers mounted in Ussing chambers as indicated. Compared to mucosal addition, serosal addition caused a reversal of the direction of the ISC induced, resulting in a drop in the basal ISC. Values are means ± the SEM of five (mucosal addition) or six (serosal addition) monolayers.
Effect of Aeromonas supernatant on ion transport of native human colon epithelium.In a limited set of experiments we tested the effect of Aeromonas sp. strain Sb supernatant on short circuit current of native human colon epithelium. As depicted in Fig. 5, after an initial ISC peak, immediately after addition of the supernatant, the Sb supernatant elicited a prolonged and marked increase in ISC. Thus, native human colon was responsive to Aeromonas sp. strain Sb supernatant as well. Apart from the initial peak, the ISC response observed in native colon epithelium was quite similar to that of HT-29/B6 monolayers.
Effect of Aeromonas supernatant on ISC of human colon epithelium. Supernatant (dilution 1:10) from an overnight culture of strain Sb was added to the mucosal compartment of human colon epithelium derived from surgical specimens and stripped from underlying subepithelial tissue. Values are means ± the SEM of five (addition of supernatant) or four (control) epithelia.
Ion and mannitol fluxes induced by Aeromonas supernatant and ion replacement studies.In order to identify the ion transport processes responsible for the ISC induced by Aeromonas supernatant, unidirectional NaCl fluxes before (control) and after mucosal addition of Aeromonas supernatant (strain Sb) were performed (Table 1). All unidirectional fluxes increased after addition of the supernatant, indicating a significant increase in paracellular permeability induced by the supernatant. Unidirectional ms and sm fluxes were used to calculate net fluxes reflecting vectorial ion transport. Under control conditions, there were no significant net fluxes. However, after addition of the supernatant, the net Cl− flux (representing chloride secretion), significantly increased, whereas sodium absorption (net Na+ flux) and bicarbonate secretion (negative residual flux) remained unchanged. Quantitatively, there was an excellent correlation between ISC (1.52 ± 0.08 μmol h−1 cm−2) and Cl− secretion (1.58 ± 0.52 μmol h−1 cm−2). The ionic basis of the ISC induced by Aeromonas supernatant was further investigated by ion replacement studies. Reducing the the Cl− concentration to 12.5% of the standard bathing solution (16.3 mM) by serial dilution with Cl−-free solution led to a 40% decrease of supernatant-induced ISC (14.9 ± 0.9 μA/cm2 versus 35.5 ± 1.5 μA/cm2 [P < 0.001, n = 6 for each group]), and subsequent elevation of the Cl− concentration within the bathing solution restored the ISC to 89% of the initial value (32.0 ± 3.3 μA/cm2, Fig. 6). Taken together, both experiments indicate that the ISC induced by the Aeromonas sp. strain Sb supernatant was mainly accounted for by electrogenic chloride secretion.
Effect of low-chloride bathing solution on Aeromonas supernatant-induced ISC. The ISC was induced by mucosal addition of Aeromonas strain Sb supernatant as described above. When submaximum ISC values (high Cl) were reached, the chloride content of the bathing solution was reduced to approximately 16 mM by repeated solution exchange. When the ISC values had stabilized after an equilibration period of 10 min (Cl reduced), chloride was added again by repeated solution exchange, and values were recorded after another 10 min of equilibration (Cl added again). Values are means ± the SEM of six monolayers. ❋❋❋, P < 0.001 versus submaximum ISC values (high Cl) recorded immediately before chloride withdrawal.
Bidirectional NaCl fluxes after addition of Aeromonas supernatant in HI-29/B6 monolayersa
In order to further characterize the drop in epithelial resistance caused by Aeromonas supernatant, experiments with labeled mannitol (a marker of paracellular permeability) were performed. After 3 h of mucosal incubation with Aeromonas supernatant, the transepithelial resistance of the monolayers had decreased to <10% of the initial value. At the same time, unidirectional mannitol fluxes revealed a 4.6-fold increase of mannitol permeation (P < 0.001), indicating a significant increment of paracellular permeability (Table 2).
Mannitol fluxes after addition of Aeromonas supernatant in HI-29/B6 monolayersa
Structural alterations of monolayers after treatment with Aeromonas supernatant.In order to test for disrupture of the epithelial barrier after damage of epithelial cells, monolayers were incubated with Aeromonas sp. strain Sb supernatant in the Ussing chamber and fixed in methanol after the transepithelial resistance had dropped to <20% of the initial value. Visualization of the tight junction protein ZO-1 showed an intact meshwork pattern with no gaps or other visible differences to controls processed in parallel (Fig. 7).
Immunofluorescence localization of ZO-1 in HT-29/B6 monolayers. (A) Monolayer after 2 h of treatment with Aeromonas strain Sb supernatant (1:10); (B) controls treated for 2 h with carrier only.
Furthermore, in independent experiments, LDH release from Aeromonas supernatant-treated HT-29/B6 monolayers was determined again, after the Rt had dropped to <20% of the initial value. Since the percentage of LDH released into the supernatants was equal in controls and Aeromonas supernatant-treated cells (1.6% ± 0.2% versus 1.2% ± 0.1% [not significant, n = 6 for each group]), the supernatant induced decrease in Rt was obviously not a consequence of necrosis of epithelial cells.
Signal transduction of Aeromonas sp. strain Sb supernatant-induced ISC.Since it has been shown that protein kinase A, protein kinase C, and elevation of intracellular Ca2+ comprise the most important stimulatory signal transduction pathways of chloride secretion (5, 13), inhibitors of protein kinase A (H8), protein kinase C (chelerythrine), and a membrane-permeable Ca2+ chelator (BAPTA-AM) were tested for their inhibitory action on the ISC induced by Aeromonas supernatant, as was a tyrosine kinase inhibitor (genistein). Prior to these experiments, all inhibitors were tested with respect to their action on HT-29/B6 cells. Thus, H8 inhibited the stimulatory effect of the cyclic AMP mobilizer forskolin (1 μM) on ion secretion of HT-29/B6 cells, chelerythrine inhibited ion secretion induced by the phorbol ester phorbol myristate acetate (5 nM), and BAPTA-AM inhibited the ion secretion elicited by the Ca2+-dependent agonist carbachol (100 μM). Since a tyrosine kinase-dependent agonist inducing chloride secretion in HT29-B6 cells is not available, genistein could not be tested in the same way as the other inhibitors. However, it was shown previously that genistein inhibited the drop in Rt induced by tumor necrosis factor alpha in HT-29/B6 cells (38). Thus, we included genistein in our experiments in order to investigate the involvement of tyrosine phosphorylation in the barrier effect elicited by the Aeromonas supernatant. For quantitative comparison, the ISC induced by Aeromonas supernatant after preincubation with the respective inhibitor was related to the ISC induced by Aeromonas supernatant without inhibitor as determined in parallel experiments. As shown in Fig. 8, inhibition of protein kinase C by chelerythrine significantly inhibited the chloride secretion induced by strain Sb supernatant (ISC induced 32 ± 2 μA/cm2 with chelerythrine versus 42 ± 2 μA/cm2 without chelerythrine [n = 6, P < 0.005]), whereas inhibition of protein kinase A or tyrosine kinase, as well as intracellular Ca2+ chelation, had no effect. In contrast to the ISC, the drop in Rt was not significantly affected by any of the inhibitors used (data not shown). Taken together, activation of protein kinase C obviously plays a role in the secretory action of the Aeromonas sp. strain Sb supernatant, whereas protein kinase A and intracellular Ca2+, which are commonly regarded as the most important regulatory signaling pathways of epithelial chloride secretion, do not seem to be involved.
Effect of inhibitors of Aeromonas supernatant induced ISC. H8 (50 μM) is a protein kinase A inhibitor, chelerythrine (10 μM) is a protein kinase C inhibitor, genistein (185 μM) acts as a tyrosine kinase inhibitor, and BAPTA-AM (30 μM) is a membrane-permeable Ca2+ chelator. The inhibitors were given 30 min prior to the addition of the supernatant. The maximum supernatant (1:10)-induced ISC in the presence of the respective inhibitor was compared to the maximum supernatant-induced ISC in monolayers pretreated with carrier only (set as 100%). Values are means ± the SEM of five or six monolayers for each group.
Selected physicochemical properties of the putative Aeromonas toxin secreted into the supernatant.For a size estimate of the active compound, aliquots of Aeromonas supernatant were ultrafiltrated by using filters with different exclusion sizes. The potential of the different fractions to induce chloride secretion and to impair the transepithelial resistance of HT-29/B6 monolayers was then investigated in the Ussing chamber. As shown in Fig. 9, the size of the active compound could be narrowed down to between 30 and 100 kDa.
Effect of size-fractionated Aeromonas Sb supernatant on the ISC value. The <30-kDa fraction was inactive, whereas the <100-kDa fraction elicited the same ISC as the concentrate of the <30-kDa fraction retained by the filter (concentrate). Thus, the active toxin secreted by strain Sb has a molecular mass between 30 and 100 kDa. Control values were obtained with unfiltrated supernatant. Values are means ± the SEM of six monolayers for each fraction.
Since both heat-labile and heat-stable exotoxins have been described for Aeromonas, the activity of the supernatant was tested for heat sensitivity. After mild heat treatment (10 min at 56°C), the effects of the Aeromonas supernatant on ISC and transepithelial resistance were completely abolished (i.e., an ISC increase after the addition of untreated supernatant of 26.8 ± 0.7 μA/cm2 versus −0.43 ± 0.7 μA/cm2 after the addition of heat-inactivated supernatant). Heat sensitivity (56°C, 10 min) has been described for Aeromonas beta-hemolysin, a 50-kDa toxin actively secreted by many Aeromonas strains and considered to be the single most important virulence factor of Aeromonas (3). Interestingly, when grown on blood agar plates, only pathogenic strain Sb but not apathogenic strain Ha produced broad zones of beta-hemolysis, and the supernatant of Sb cultures grown overnight were positive in the hemolysin titer test (between 8 and 16 hemolysin units) compared to no measurable activity for the Ha supernatant. It was therefore hypothesized that the epithelial effects observed were caused by beta-hemolysin secreted by pathogenic aeromonads. Since zinc has been reported to inhibit oligomerization and channel function of this pore-forming toxin (40), we performed inhibition experiments with zinc. Zinc indeed significantly inhibited the effects of Aeromonas sp. strain Sb supernatant on ion secretion and barrier function of HT-29/B6 monolayers (Fig. 10) since, in the presence of zinc, maximum supernatant-induced ISC amounted to <50% of the ISC obtained without zinc in parallel experiments (supernatant-induced ISC of 42.1 ± 1.7 μA/cm2 in the presence of zinc versus an ISC of 97.2 ± 6.8 μA/cm2 without zinc).
Inhibitory effect of ZnCl2 on Aeromonas Sb supernatant-induced ISC and resistance decrease (Rt). ZnCl2 was added to the Aeromonas strain Sb supernatant to a final concentration of 5 mM; Aeromonas strain Sb supernatant without ZnCl2 was used for controls. The experiment was started by filling the mucosal compartment with undiluted supernatant with or without ZnCl2. Zinc clearly inhibited the supernatant-induced ISC by >50% (A) and the supernatant-induced resistance decrease (B). The Rt was determined for comparison 30 min after addition of the supernatant. Values are means ± the SEM of six monolayers for each group. ❋❋❋, P < 0.001 versus controls treated with supernatant only.
Effects of recombinant Aeromonas beta-hemolysin on ion secretion and barrier function of HT-29/B6 monolayers.In order to obtain conclusive evidence that Aeromonas beta-hemolysin can cause intestinal ion secretion and barrier impairment, we cloned the beta-hemolysin gene from strain Sb. The complete coding sequence can be retrieved from GenBank (accession number AY611033 ). Sequencing of the gene revealed a high homology to known Aeromonas beta-hemolysin sequences. The highest homology (98% identity at the amino acid level) was found in relation to Aeromonas sobria beta-hemolysin, which has been shown to cause fluid accumulation in the mouse intestinal loop assay without cellular damage as visualized by histopathologic examination (16) (GenBank accession number AY157998 .1). Less but still high homology was found with respect to the sequence of aerolysin, an Aeromonas beta-hemolysin sequenced in the late 1980s as the first of the Aeromonas beta-hemolysins. Thus, the Aeromonas strain Sb beta-hemolysin showed 68% (20) (GenBank accession number M16495 ) and 60% (21) (GenBank accession number Y00559 ) amino acid identity to the aerolysin precursor (preproaerolysin) and 71% (20) and 64% (21) identity to the mature aerolysin that emerges from preproaerolysin after intracellular and extracellular cleavage of N- and C-terminal peptides (21).
To functionally characterize the Aeromonas strain Sb beta-hemolysin, cell lysates from overnight cultures of E. coli TOP10F′ transformed with the beta-hemolysin gene from strain Sb were taken for electrophysiological experiments. Lysates from TOP10F′ cells transformed with the vector alone served as controls. For comparison, experiments with transformed DH5α cells with or without the beta-hemolysin (aerolysin) gene previously identified (21) were performed in an analogous manner. As shown in Fig. 11A, cell lysate of untransformed TOP10F′ cells did not exert any effect on the transport and barrier properties of HT-29/B6 cells. If, however, the TOP10F′ cells contained the Aeromonas beta-hemolysin gene, the cell lysates evoked a strong secretory response, as well as a rapid decline of transepithelial resistance. Identical results were obtained with DH5α cells transformed with or without the Aeromonas beta-hemolysin (Fig. 11B). As expected, the cell lysates exerted strong hemolytic activity with 16 hemolytic units for TOP10F′ cells harboring the Aeromonas beta-hemolysin gene and 256 hemolytic units for DH5α cells harboring the aerolysin gene. Lysates of TOP10F′ or DH5α cells transformed with the respective vectors alone did not show any hemolytic activity in the titer test (data not shown).
Effect of cloned Aeromonas beta-hemolysins on ISC and Rt of HT-29/B6 monolayers. (A) Effect of cell lysate from untransformed TOP10F′ cells (control, dotted line) or TOP10F′ cells transformed with the Aeromonas Sb beta-hemolysin gene (Sbhls) on ISC (upper panel) and Rt (lower panel) of HT-29/B6 monolayers; (B) effect of cell lysate from untransformed DH5α cells (control, dotted line) or DH5α cells transformed with a beta-hemolysin gene (aerolysin) derived from A. trota (DH5aerA, solid line) on ISC (upper panel) and Rt (lower panel) of HT-29/B6 monolayers. Values are means ± the SEM of six monolayers for each group.
DISCUSSION
Data from numerous epidemiological, clinical, and experimental studies provide strong evidence that pathogenic aeromonads can cause diarrheal disease. However, how they do this is not known, i.e., it is has not been determined whether intestinal fluid secretion in Aeromonas enteritis is only caused by an impaired epithelial barrier function or whether active ion secretion is also activated. Furthermore, there is little information about the role of Aeromonas virulence factors in this context. This lack of information is due to the heterogeneity of the genus Aeromonas, the great variety of Aeromonas virulence factors, and the loose association between virulence factors and the phenotype and genotype (3, 22, 23). For these reasons, a straightforward correlation with a single virulence factor and the pathogenic effects of Aeromonas enteritis has not been demonstrated thus far. As a consequence, many researchers have proposed a multifactorial pathogenesis with the involvement of a number of cellular and extracellular virulence factors (9, 19).
In order to study the events leading to intestinal fluid secretion in enteral Aeromonas infection, we established a cell culture infection model that allows investigation of the epithelial effects of aeromonads (and other intestinal pathogens) with high resolution. Although the findings obtained with this model do not, of course, reflect all aspects of the pathology of intestinal Aeromonas infections, this approach offers some important advantages. The basic setting can be seen as a reductionistic approximation with an intestinal epithelium exposed to a luminal (i.e., mucosal) pathogen. Therefore, chances are good to detect effects resulting from the interplay of different virulence factors expressed from the same pathogen. In addition, differentiation between cellular (e.g., adherence and invasiveness) and extracellular virulence factors secreted from the bacteria can easily be achieved. Furthermore, our cell culture infection model allows real-time observation of functional epithelial effects elicited by enteropathogens, and interpretation of these effects is straightforward compared to that of pathogenic effects observed in more complex organ model or in vivo systems. Finally, the biological significance of effects observed in our cell culture model was corroborated by experiments with native human colonic epithelium that showed a qualitatively similar response to Aeromonas supernatant such as the monolayers.
The first important finding of our study was that a pathogenic Aeromonas strain was capable of inducing both active ion secretion and significant barrier impairment. Although adherence and invasiveness are known virulence factors of certain Aeromonas strains (22, 25, 33), a direct interaction between bacteria and epithelium was obviously not a prerequisite for the effects of strain Sb on ion secretion and barrier function. On the contrary, all epithelial effects observed in our model could likewise be induced by sterile Aeromonas supernatant. Moreover, several lines of evidence leave no doubt that the effects elicited by Aeromonas strain Sb supernatant were caused by a secretory beta-hemolysin. First, we observed a correlation between the capacity of the Aeromonas supernatant to induce ion secretion and barrier impairment and its hemolytic activity, whereas an apathogenic strain with no hemolytic activity in the titer test did not exert any effects on ion secretion or barrier properties of the HT-29/B6 monolayers. Furthermore, considering known Aeromonas virulence factors, the physicochemical properties (molecular weight, heat sensitivity, and inhibition by zinc) of the supernatant's active compound leave an aerolysin-related beta-hemolysin as the main suspect and, most strikingly, the beta-hemolysin cloned from Aeromonas strain Sb and expressed in E. coli exerted identical effects with the HT-29/B6 cells as with supernatant from Aeromonas strain Sb. Finally, the beta-hemolysin obtained from A. hydrophila strain Sb was highly homologous to other aerolysin-related Aeromonas beta-hemolysins, and the aerolysin cloned previously from an Aeromonas trota strain (21) also caused effects in our experiments identical to those of the beta-hemolysin of strain Sb. Taken together, our data prove that aerolysin and related Aeromonas beta-hemolysins induce active chloride secretion and significant barrier impairment in the intestinal epithelium. Although adherence to the epithelium may play a role in vivo by increasing the local Aeromonas density, the beta-hemolysin obviously sufficed to elicit the epithelial effects observed without the involvement of additional virulence factors.
This finding underlines the significance of the aerolysin family of beta-hemolysins, which have been suspected for a long time to be the major virulence factor of Aeromonas enteritis (9). Aerolysin is a pore-forming toxin first identified by Bernheimer and Avigad in 1974 (6). Later, a number of Aeromonas beta-hemolysins with similar properties (molecular weight, heat sensitivity, etc.) were identified and now more than 10 full-length sequences of aerolysin-related Aeromonas beta-hemolysins can be retrieved from GenBank that share between 30 and 99% homology with each other. A great diversity of pathogenic effects have been described for these beta-hemolysins, such as fluid accumulation in intestinal loops (4, 14), release of inflammatory mediators from granulocytes (37), and induction of apoptosis in human lymphoma cells (32). The data explaining the mechanism of the putative diarrheal action of this toxin are sparse, however. Only recently, one study was published investigating the interaction of aerolysin with its primary target in enteric disease: the intestinal epithelium (2). In this case, aerolysin caused a dose-dependent drop in transepithelial resistance, indicating barrier impairment by aerolysin. Our results confirm and extend the findings of that study.
According to our data, both passive solute and water secretion secondary to barrier impairment and active chloride secretion contribute to the fluid secretion in Aeromonas-induced diarrhea. Thus, we demonstrate for the first time that aerolysin-related Aeromonas beta-hemolysins not only cause a decrease of transepithelial resistance but also induce active chloride secretion. This finding came as a surprise because many researchers interpreted the diarrheic action to result from intestinal barrier damage after the killing of epithelial cells by membrane permeabilization (15, 35). However, despite the marked effects on epithelial permeability, we did not find any disrupture of the epithelial monolayer, as visualized by ZO-1 immunofluorescence staining. Furthermore, LDH release from the epithelial cells was not enhanced by treatment with Aeromonas supernatant. These results are in line with other studies showing that, depending on the toxin concentration, mammalian cells survive several hours of aerolysin treatment (1) and that mouse intestinal loops treated with an Aeromonas beta-hemolysin closely related to the hemolysin obtained from strain Sb (98% sequence homology at amino acid level) developed significant fluid secretion without histopathologic evidence for epithelial cell damage or mucosal inflammation (16). Furthermore, this Aeromonas beta-hemolysin stimulated cyclic AMP production in T84 cells without causing cell death, as evidenced by LDH release assay and trypan blue exclusion (17), and aerolysin triggered G-protein activation and Ca2+ release from intracellular stores but not cell lysis in human granulocytes (26). Taken together, there is strong evidence that Aeromonas beta-hemolysins of the aerolysin family can trigger complex cellular events independently from cell lysis. This concept is corroborated by demonstration of active chloride secretion induced by Aeromonas beta-hemolysin, which cannot be explained by cell death.
However, what precisely is the particular mechanism of Aeromonas beta-hemolysin induced chloride secretion? The general scheme of chloride secretion in intestinal epithelial cells involves basolateral chloride entry by the Na+/K+/2Cl− symporter NKCC1, potassium recycling via basolateral potassium channels, and apical chloride outflow via apical chloride channels like, e.g., the cystic fibrosis transmembrane conductance regulator (CFTR) (5). The energy is supplied by the Na+/K+-ATPase, which promotes sodium coupled chloride entry by maintaining a low intracellular sodium concentration. The apical chloride outflow is generally considered the rate-limiting step, and activation of protein kinase A and elevation of intracellular Ca2+ have been identified as the most prominent stimulating pathways (5). In addition, it has been shown that protein kinase C and tyrosine kinases are also potential activators of apical chloride channels (5). In our study, only the protein kinase C inhibitor chelerythrine inhibited Aeromonas beta-hemolysin induced chloride secretion. Therefore, it is possible that activation of protein kinase C is involved in the secretory action of the beta-hemolysin, although we did not explore the signal transduction involved in this stimulation. In searching for a potential link between protein kinase C and apical chloride channels, we became aware of a chelerythrine-sensitive, calcium-independent protein kinase C activation of the CFTR in polarized airway epithelium in response to apical ATP (via purinergic receptors) (34). Thus, protein kinase C activation of CFTR could occur also in HT-29/B6 cells in response to Aeromonas. The initial event would then be apical release of ATP, as recently described for an aerolysin-like hemolysin from A. sobria in T84 cells (18).
However, since chelerythrine only inhibited a minor part of the overall secretion, the major pathway of Aeromonas hemolysin-induced chloride secretion seems to be independent from protein kinase C, as well as from protein kinase A or calcium. According to extensive experimental evidence in the literature, Aeromonas beta-hemolysin effects are generally thought to follow pore formation by insertion of aerolysin oligomers into the membrane of the target cell (see, for example, references 2, 8, 14, 26, and 32; for reviews, see references 15 and 35). Even intracellular events triggered by Aeromonas beta-hemolysin have been assumed to occur secondary to formation of the hemolysin pore within the cell membrane of the target cell (1, 15). Therefore, we hypothesize that the effects observed in our cell model were also caused by membrane insertion of Aeromonas hemolysin. Furthermore, it has been shown in patch clamp studies that the pore formed by the Aeromonas beta-hemolysin is an unselective ion channel with a slight preference for anions such as chloride (8, 40). When the basolateral components of epithelial chloride secretion are constitutively active, insertion of an apical chloride channel completely suffices to elicit active chloride secretion. From these findings, it seems likely that, once the aerolysin pore has been inserted into the apical membrane, Cl− leaves the cell along its electrochemical gradient via the hemolysin pore, leading to active Cl− secretion.
As for the inhibitory effect on ISC of Aeromonas hemolysin applied serosally, the reasoning presented above and the velocity of the inhibition strongly support the notion that this effect may be caused by basolateral insertion of the hemolysin pore. Since the aerolysin pore is an unselective ion channel, the cessation of all electrogenic ion movement can easily be explained by the breakdown of the basolateral membrane potential abolishing the driving force for all active ion transport processes. We propose, on the basis of these results, that chloride secretion induced by aerolysin-related beta-hemolysins involves protein kinase C activation, as well as chloride outflow through the ion pores formed by the hemolysin inserted into the apical membrane, although it should be noted that direct evidence for the latter hypothesis of functionally active channel insertion into the apical membrane is difficult, if not impossible, to obtain in this experimental system.
Regarding the effect of Aeromonas on the epithelial barrier, we observed a marked decrease in transepithelial resistance induced by Aeromonas beta-hemolysins similar to the effect of aerolysin on Caco-2 cells described by Abrami et al. (2). Thus, evidence is mounting that beta-hemolysins of the aerolysin family can cause severe impairment of the intestinal epithelial barrier. Although the aerolysin pore itself may contribute to this process to some degree, our flux data indicate that an increase in paracellular rather than transcellular permeability is the main correlate for the resistance decrease observed. In principle, an increment of paracellular permeability can occur either after disintegration of the epithelial monolayers after the death of epithelial cells or after structural alteration of the paracellular seal, the tight junction. Although there is little doubt that aerolysin, depending on the locally active concentration, can cause cell death by osmotic lysis (1), our data, surprisingly, seem to point to alteration of the tight junction and away from cell necrosis as the main mechanism of the aerolysin-induced barrier impairment observed in our experiments, since with the aerolysin concentration and the incubation time used in our experiments we did not find any evidence of death of HT-29/B6 cells. In agreement with the results of others (17), there was no increase in LDH release after hemolysin treatment, and, morphologically, Aeromonas hemolysin-treated monolayers did not show any leaks or disruptions of the epithelial monolayer. Furthermore, the resistance values did not decrease completely but remained stable at a low level for a prolonged period of time. However, a direct effect on tight-junction strand-forming proteins, such as occludin, claudin-1, and claudin-4, was not quantitatively investigated in the present study.
In summary, we have demonstrated for the first time that beta-hemolysins of the aerolysin family not only induce impairment of the intestinal epithelial barrier but also stimulate active chloride secretion. Although our data are compatible with chloride secretion via the hemolysin pore inserted into the apical membrane and to some extent by protein kinase C activation, the hemolysin-induced barrier impairment seems to be due to an indirect hemolysin-mediated increase in paracellular permeability.
ACKNOWLEDGMENTS
This study was supported by grants from Deutsche Forschungsgemeinschaft (DFG Schu 559/7-3 and DFG KFO 104/1).
The excellent technical assistance of Anja Fromm, Friederike Hirsch, and Ursula Lempart is gratefully acknowledged. We thank Eugen Domann, Department for Medical Microbiology, University of Giessen, for taxonomic classification of the strains by molecular typing.
FOOTNOTES
- Received 18 November 2003.
- Returned for modification 5 February 2004.
- Accepted 2 April 2004.
- Copyright © 2004 American Society for Microbiology
