In Vitro Effects of a High-Molecular-Weight Heat-Labile Enterotoxin from Enteroaggregative Escherichia coli

Abstract

The pathogenic mechanisms of enteroaggregative Escherichia coli (EAggEC) infection are not fully elucidated. In this work we show that an ammonium sulfate precipitate of culture supernatant of EAggEC strain 049766 increased the potential difference (PD) and the short-circuit current (Isc) in rat jejunal preparations mounted in Ussing chambers. The precipitate contained two major proteins of 108 and 116 kDa, which were partially copurified by chromatography in DEAE-cellulose. This chromatographic fraction (peak I) increased jejunal PD and Isc in a dose-dependent manner, accompanied by a decrease in tissue electrical resistance. These effects were inhibited by incubation of peak I at 75°C for 15 min or for 1 h with proteinase K at 37°C. Rabbit polyclonal antibodies against peak I containing both the 108- and 116-kDa proteins inhibited the enterotoxic effect. Specific polyclonal antibodies raised against the 108-kDa but not against the 116-kDa protein inhibited the enterotoxic effect, suggesting that the 108-kDa protein is the active toxic species. Moreover, another EAggEC strain (065126) producing the 116-kDa protein but not the 108-kDa protein had no effect on rat jejunal mucosa in the Ussing chamber. The >100-kDa fraction derived from prototype EAggEC strain 042, which also expressed both 108- and 116-kDa proteins, also produced an enterotoxic effect on rat jejunal preparations in Ussing chambers; however, the same strain cured of its 65-MDa adherence plasmid did not. A subclone derived from the 65-MDa plasmid expressing the 108-kDa toxin (and not the 116-kDa protein) elicited rises in Isc. Tissue exposed to any preparation containing the 108-kDa toxin exhibited similar histopathologic changes, characterized by increased mucus release, exfoliation of cells, and development of crypt abscesses. Our data suggest that some EAggEC strains produce a ca. 108-kDa enterotoxin/cytotoxin which is encoded on the large virulence plasmid.

Enteroaggregative Escherichia coli (EAggEC) has been associated with persistent diarrhea in young children (3, 5, 15, 25), especially in developing countries. Most EAggEC strains harbor a 65-MDa plasmid (called pAA), which is required for expression of aggregative adherence fimbriae (AAFs). These structures mediate the defining aggregative adherence (AA) phenotype to HEp-2 cells (14, 15, 21) as well as adherence to the colonic mucosa (7). The pAA plasmid is also required for the development of mucosal damage in in vitro models (10, 16).

Clinical, volunteer, and animal model studies suggest that EAggEC diarrhea may be due to a secretogenic enterotoxin (3, 13, 25). When tested in Ussing chambers, filtrates from EAggEC strain 17-2 produced an increase in potential difference (PD) and short-circuit current (Isc), attributed by Savarino et al. (20) to a heat-stable, plasmid-encoded enterotoxin of less than 10 kDa in molecular mass (EAST1). No data yet exist to support a role for EAST1 in EAggEC diarrhea.

Nataro et al. (13) reported that EAggEC strain 042 (serotype O44:H18) caused significant diarrhea in three of five adult volunteers, whereas strains 17-2, 34b, and JM221 (EAggEC of serotypes O3:H2, O?:H?, and O92:H33, respectively) did not induce enteric symptoms. Except for 042, each of these strains expressed AAF/I fimbriae, while EAST1 was produced by strains 042 and 17-2 but not by strain 34b or JM221 (13). Mathewson et al. (12) had previously reported that strain JM221 caused mild diarrhea in some adult volunteers, and Tzipori et al. (23) had shown that strains 17-2 and JM221 were able to cause diarrhea in gnotobiotic piglets. The basis for this strain heterogeneity has not been determined, and these data suggest the presence of unrecognized virulence factors.

A role for cytotoxins in EAggEC disease has been suggested by in vivo and in vitro models (10, 16, 24), which exhibit damage to intestinal epithelium. A candidate cytotoxin has been identified by Eslava et al. (8), as sera from children in a Mexican EAggEC outbreak consistently recognized two proteins of 108 and 116 kDa obtained from ammonium sulfate precipitates of EAggEC culture supernatants. Together these proteins elicited hemorrhagic and necrotic lesions in rat ileal loops (8). In this communication, we report that a fraction containing the 108- and 116-kDa proteins purified from EAggEC 049766, the strain implicated in the Mexican outbreak, is able to cause enterotoxic and cytotoxic effects on rat jejunal tissue mounted in Ussing chambers. We have identified the toxin moiety as the 108-kDa protein and have localized it to the pAA plasmid.

MATERIALS AND METHODS

Bacterial strains.

EAggEC strain 049766, implicated in an outbreak of persistent diarrhea in Mexican infants, has been characterized as belonging to serotype O?:H10 and is capable of attaching with an aggregative pattern to HEp-2 cells (6, 15). EAggEC strain 065126 was isolated from a Mexican child with diarrhea. Strain 042 was isolated from a child with diarrhea in Lima, Peru, in 1983; this strain has been shown to cause diarrhea in adult volunteers (13). Strains 049766, 065126, and 042 hybridize with the EAggEC-specific (AA) probe (2). Plasmid-cured 042 has been described previously (16). E. coli strain K-12 was used as a control in Ussing chamber experiments; E. coli HB101 was used as a host for cloning experiments.

Preparation of protein fractions.

Culture conditions and preparation of precipitates were performed as follows. EAggEC strain 049766 was grown overnight at 37°C in 200 ml of Luria broth (LB). After centrifugation at 12,000 × g for 10 min, supernatants were precipitated with 60% saturated ammonium sulfate for 18 h at 4°C, collected by centrifugation, dissolved with 0.07 M potassium phosphate buffer (pH 8.2), and dialyzed for 4 days against the same buffer. Protein concentration was determined by the method of Bradford (4). Purification of EAggEC-secreted proteins was obtained by further precipitation of protein suspensions in 0.07 M potassium phosphate buffer with 1.75 M K₂HPO₄, dialyzed at 4°C against Tris-EDTA buffer (0.05 M Tris-0.1 M EDTA [pH 8.0]) and eluted with the same buffer from a DEAE-cellulose column.

For the neutralization experiments in the Ussing chamber, performed at the Center for Vaccine Development at Baltimore, EAggEC strain 049766 was grown overnight at 37°C in 100 ml of LB. After centrifugation at 12,000 × g for 10 min, supernatants were concentrated and size fractionated by passage through Biomax-50 Ultrafree filters (Millipore, Bedford, Mass.) according to the manufacturer’s instructions.

Protein electrophoresis and immunologic methods.

Proteins present in precipitated culture supernatants from the DEAE chromatography fractions were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) by the method of Laemmli (11) under reducing conditions (boiling for 5 min in the presence of mercaptoethanol). Proteins separated by SDS-PAGE were transferred to nitrocellulose BA85 membranes (Schleicher & Schuell, Keene, N.H.) by the method of Towbin et al. (22). The membranes were incubated with rabbit antibodies generated in our laboratory against the 108- and 116-kDa proteins (diluted 1:100) or with sera from children in the Mexican EAggEC outbreak (diluted 1:25). Immunostaining was performed with peroxidase-labeled polyclonal antibodies against rabbit immunoglobulins (dilution 1:5,000) and developed with 4-chloronaphthol by standard methods (22). The antibodies against each of the 108- and 116-kDa proteins were elicited by excising proteins from polyacrylamide gels and injecting the gel slices subcutaneously into rabbits in two doses, 2 weeks apart. The antibody responses and specificity were determined by immunoblotting. The gamma fractions from the antisera were diluted 1:25 prior to use in Ussing chamber experiments.

Electrophysiologic measurements in rat jejunum.

Ussing chamber experiments in Mexico City were performed as described previously (18, 19). Jejunal segments removed from adult male Sprague-Dawley rats under sodium pentobarbital anesthesia were placed in ice-cold Ringer’s solution for mammals and gassed with an O₂-CO₂ (95%:5%) mixture. The excised segments were cut open along the mesenteric border, washed with cold Ringer’s solution, divided into two fragments (experimental and control), and mounted between the circular openings (6-mm diameter, 0.28 cm²) of two adjacent Ussing hemichambers. Each hemichamber was filled with 10 ml of Ringer’s solution and kept at 37°C under constant O₂-CO₂ bubbling. Transmural PD and Isc were recorded at 1-min intervals by means of a voltage clamp apparatus. Samples containing 1.5 to 25 μg of precipitate per ml were diluted with Ringer’s solution at 37°C and added to the mucosal hemichamber of rat jejunum preparations after 10 min of equilibration, and both hemichambers were gassed with O₂-CO₂. Transmural resistance (R) values were obtained from PD and Isc values by using Ohm’s law. Statistical analyses were performed with Student’s t test on data recorded from at least four experiments.

Ussing chamber experiments in which enterotoxic activity was inhibited by antibodies against either the 108- or 116-kDa protein were performed at the University of Maryland by methods previously described (9). Six pieces of rat jejunum were mounted in Ussing chambers; a known positive control and appropriate negative control were always assayed in parallel with the test samples (culture filtrates of strain 049766 with or without antibodies), using the same rat tissue. PD was measured at intervals, and total tissue conductance and Isc were calculated (9).

Supernatants from strains 049766, 042, 065126, and HB101(pJPN201) used in Ussing chamber experiments were concentrated 100× and size fractionated by using Biomax Ultrafree filters (100-kDa cutoff; Millipore) according to the manufacturer’s instructions. Neutralization of the electrophysiological effects of EAggEC proteins was tested by using aliquots containing 25 μg of partially purified EAggEC proteins that were either heat treated at 75°C for 15 min or incubated with proteinase K (200 μg/ml) at 37°C for 1 h before being added to the luminal side of jejunal preparations mounted in Ussing chambers. To test the inhibitory effects of different antibodies, 25 μg of partially purified EAggEC proteins was preincubated for 20 min at room temperature with rabbit polyclonal antibodies directed against the 108- or 116-kDa protein (diluted 1:25) before being added to the luminal hemichamber.

To assess the integrity of intestinal preparations at the end of the electrophysiologic experiments, samples were removed from Ussing chambers and fixed for 1 h in 10% formalin, embedded in paraffin, and cut into 4- to 6-μm sections that were stained with hematoxylin and eosin (18) and examined under light microscopy.

RESULTS

Ussing chamber effects of 108- and 116-kDa supernatant proteins.

Precipitation of EAggEC 049766 culture supernatants using 60% saturated (NH₄)SO₄ yielded several proteins, most prominent of which were 108- and 116-kDa species (Fig. 1, lane C) that were absent from precipitates of culture supernatants of an E. coli K-12 strain (lane A). Addition of the precipitates obtained from the 60% saturated (NH₄)₂SO₄ supernatant of 049766 (lane C) to the mucosal hemichamber of rat jejunum strips mounted in Ussing chambers evoked a significant increase of PD and Isc while producing a decrease of R (Fig. 2). Rises in PD and Isc began approximately 20 min after addition of culture supernatants. PD and Isc rose from 0.5 to 1.04 mV and from 5.1 to 16.7 μA/cm², respectively, while the R values decreased about 50% (P = 2.8 × 10⁻⁸), from 102 to 58 Ωcm². Maximum increases were attained approximately 95 min after inoculation. Precipitates from culture supernatants of an E. coli K-12 and from uninoculated broth had no effect on jejunal preparations from the same animals (Fig. 2). These data suggest that the supernatant of strain 049766 contains an enterotoxin.

FIG. 1.

SDS-PAGE characterization of protein fractions (40 μg per lane) from EAggEC strain 049766. E. coli K-12 was used as a control. The crude 60% (NH₄)₂SO₄-precipitated supernatant (lane C) of 049766 produced several proteins, including those at 108 and 116 kDa. The high-molecular-weight fraction was partially purified by reprecipitation with 1.75 M K₂HPO₄ (lane B) and chromatography through DEAE-cellulose (peak I, lane D). Concentrated supernatant of E. coli K-12 (lane A) did not reveal secreted proteins in the range of 108 to 116 kDa.

FIG. 2.

Time course of PD, Isc, and R values of rat jejunum preparations exposed to 60% (NH₄)₂SO₄-precipitated supernatants from cultures of EAggEC strain 049766 or E. coli K-12. Twenty-five micrograms of protein was used from concentrated supernatants or 25 μl of uninoculated LB. The symbols represent the mean values of experiments performed on four different animals.

To identify the protein conferring the enterotoxic effect, 049766 precipitated supernatant preparations were enriched by a second precipitation with 1.75 M K₂HPO₄ (Fig. 1, lane B) and then were separated by DEAE-cellulose chromatography (lane D). The first peak obtained from the DEAE-cellulose column (hereafter designated peak I) produced a highly enriched fraction containing both the 108- and 116-kDa EAggEC proteins.

The peak I supernatant fraction induced increases in rat jejunal PD and Isc, and affected R, similarly to the crude 049766 precipitates. The effect of peak I proteins on Isc values was dose dependent (1.5 μg of protein induced mean increases in Isc of 2.07 μA/cm², while 25 μg induced mean Isc rises of 13.06 μA/cm²), starting 20 min after the inoculation (Table 1). In addition, the same mass of protein (25 μg) from the peak I fraction induced a greater increase of PD and Isc than the crude precipitate (0.71 mV and 14.32 μA/cm² [peak I] versus 0.44 mV and 10.6 μA/cm² [crude]). These data strongly suggest that the 108- and/or the 116-kDa proteins exhibit dose-dependent enterotoxic properties. Interestingly, however, the crude precipitate produced a significantly greater change in resistance (ΔR = 43.2 Ωcm² versus 25.06 Ωcm²), suggesting that another factor(s) may also contribute to mucosal damage.

TABLE 1.

Increase in PD and Isc after addition of various doses of partially purified 108- and 116-kDa EAggEC-secreted proteins in rat jejunum strips mounted in Ussing chambers^a

Dose (μg)	ΔPD (mV)	ΔIsc (μA/cm2)	Time of maximum increase (min)b
1.5	0.22	2.07	22
2.5	0.36	3.83	33
3.75	0.26	5.25	39
25.0	0.78	13.06	95

^aAverage values for each dose (n = 4). 

^bPD and Isc started to increase 20 min after addition of protein preparations.  

Heat-treated peak I proteins (75°C for 15 min) lost enterotoxic activity (Table 2). Preincubation with proteinase K also inhibited the effects of peak I proteins on jejunal PD and Isc (Table 2). These data are consistent with the presence of a high-molecular-weight heat-labile protein enterotoxin.

TABLE 2.

Jejunal PD and Isc values after addition of the 108- and 116-kDa EAggEC proteins preheated or preincubated with proteinase K^a

Toxin prepn	Treatment	PD (mV)	Isc (μA/cm2)
None	None	0.92	3.89
Peak I proteins	None	2.00	15.3
	Heatb	0.50	4.49
	Proteinase Kc	0.98	3.63

^aAverage values recorded 30 min after toxin addition (n = 4). 

^b25 μg of partially purified proteins heated at 75°C for 15 min. 

^c25 μg of partially purified proteins preincubated with 200 μg of proteinase K per ml for 1 h at 37°C. 

Association of enterotoxic activity with the 108-kDa protein.

The 108- and 116-kDa proteins were found to be immunogenic. Serum samples from children with diarrhea due to strain 049766 in the Mexican outbreak, reacted against the supernatant of the same strain (Fig. 3A, lane a) by Western immunoblotting, recognized either both the 108- and 116-kDa proteins (Fig. 3B, lane e) or only the 108-kDa species (Fig. 3B, lane f).

FIG. 3.

SDS-PAGE (A) and Western immunoblotting (B to D) of >100-kDa fractions from supernatants of strains 049766 (lanes a), 065126 (lanes b), 042 (lanes c), and HB101(pJPN201) (lanes d). In panel B, Western blots in lanes a to d are reacted with anti-peak I antibodies, and those in lanes e and f are reacted with antibodies from two different patients in the 049766 outbreak. Blots in panel C are reacted with anti-108-kDa protein antibodies, and those in panel D are reacted with anti-116-kDa protein antibodies. Lower-molecular-weight bands in all lanes most likely represent breakdown products of the high-molecular-weight species, since they are generally absent in blots lacking reactivity in the region from 108 to 116 kDa. MW, molecular weight markers.

We took advantage of the immunogenicity of these proteins to identify the toxic species. Rabbits immunized with peak I proteins from strain 049766 produced antibodies against both the 108- and 116-kDa proteins (Fig. 3B, lane a). Monospecific polyclonal antibodies against either the 108-kDa (Fig. 3C, lane a) or 116-kDa (Fig. 3D, lane a) protein were prepared by excising the proteins from polyacrylamide gels and injecting the proteins into different rabbits. Each of these antibody preparations was then tested for the ability to inhibit the enterotoxicity of fractionated EAggEC supernatants in the Ussing chamber. As expected, anti-peak I antibodies neutralized PD, Isc, and R changes in Ussing chambers (Fig. 4). No rises in Isc or decreases in R were detected (Fig. 4A and C). Preincubation of the peak I fraction with monospecific antibodies against the 108-kDa but not against the 116-kDa protein neutralized the effects of the preparation on jejunal PD and Isc (Fig. 5). These data suggest that the 108-kDa protein is the enterotoxic species found in the peak I fraction.

FIG. 4.

Inhibition of enterotoxicity by antibodies against the peak I fraction. Twenty-five-microgram aliquots of peak I proteins were preincubated for 20 min with rabbit serum directed against the identical fraction and then added to the mucosal hemichambers of rat jejunum preparations (n = 4).

FIG. 5.

Inhibition experiments in Ussing chambers with antibodies against either 108- or 116-kDa protein. Bars 1 to 3, PD and Isc increments induced by the peak I fraction of strain 049766 alone or after preincubation with monospecific antibodies against either the 108- or 116-kDa protein (n = 7). Bar 4 represents the rises induced by strain 065126, which lacks the 108-kDa species.

We used our 108- and 116-kDa protein-specific polyclonal antibodies to screen our collection for strains that might express only the 108- or 116-kDa protein to further support our hypothesis that the 108-kDa protein was the active species. By Western immunoblotting strain 065126 was found to express the 116-kDa (Fig. 3D, lane b) but not the 108-kDa (Fig. 3C, lane b) protein. As predicted, the >100-kDa fraction of 065126 (Fig. 3A, lane b) did not induce changes in jejunal PD and Isc (Fig. 5) and was not significantly different from the preparation treated with LB medium (P = 0.1).

Localization of the gene encoding the 108-kDa enterotoxin.

Genetic analyses in our laboratories has focused on EAggEC strain 042 (7, 24). We decided to use these data to localize the 108-kDa toxin and to substantiate its enterotoxic effects. Concentrated supernatants of strain 042 were found to contain the 108- and 116-kDa proteins, detected by SDS-PAGE (Fig. 3A, lane c) and by immunoblotting with antibodies against 108- and 116-kDa proteins (Fig. 3B, lane c). As expected from previous experiments, these concentrated supernatants also induced increases of PD and Isc (Fig. 6); however, strain 042 cured of its 65-MDa virulence plasmid (pAA2) was found to be lacking the 108-kDa protein, and the fractionated supernatant of plasmid-cured 042 had no effect on jejunal preparations mounted in the Ussing chamber (Fig. 6). We next tested a series of clones derived from plasmid pAA2 and found that HB101(pJPN201), harboring a 13-kb insert which flanks the previously described AAF/II genes (7), expressed the 108-kDa protein by SDS-PAGE (Fig. 3A, lane d) and by Western blotting (Fig. 3B and C, lanes d). The 116-kDa protein was not encoded by pJPN201 (Fig. 3D, lane d). Again, as expected, concentrated fractionated supernatant of HB101(pJPN201) induced rises in jejunal PD and Isc in Ussing chambers (Fig. 6); the rises were neutralized by anti-108-kDa protein antibodies. These data confirm that the 108-kDa protein is indeed an enterotoxin.

FIG. 6.

Enterotoxic activity of >100-kDa fractionated supernatants containing the 108-kDa protein. One-hundred micrograms of concentrated supernatant protein was added to the mucosal hemichambers of rat jejunum preparations (n = 4) (see text).

Histopathologic examination of rat mucosal tissue.

Since the 108-kDa protein induced a decrease in electrical resistance, histopathologic analysis of full-thickness rat jejunal tissue was performed by light microscopy after Ussing chamber experiments. Control-treated rat jejunal sections appeared normal, with intact mucosa and minimal mucus secretion (Fig. 7A). However, specimens treated with the 108-kDa toxin derived either from 049766 or from 042 [in HB101(pJPN201)] demonstrated identical histopathologic abnormalities (Fig. 7B). The mucosal surface of toxin-treated specimens was covered with a thick mucus blanket. The epithelial layer demonstrated coagulation necrosis, with exfoliation of epithelial cells and occasional karyorrhexis of nuclei. Beneath the epithelium were observed increased numbers of mononuclear cells, and eosinophils and multifocal crypt abscesses were observed in several specimens. The submucosa exhibited edema and widening of the lymphatic channels.

FIG. 7.

Morphologic effects of 108-kDa protein on rat jejunal mucosa. The rat jejunal preparations were removed from Ussing chambers, fixed with 4% formalin, and embedded in paraffin. The sections were stained with hematoxylin and eosin. (A) Untreated control preparation. (B) Preparation treated with 108-kDa protein from HB101(pJPN201). Note the mucus blanket with cell debris on the luminal side (asterisk), damage of the epithelial layer (arrowhead), and crypt abscesses (arrow) in the treated section.

DISCUSSION

EAggEC is an emerging agent of pediatric diarrhea. Clinically, the disease presents as watery diarrhea, but the responsible enterotoxin has not yet been identified with certainty. Data do not yet exist to support a role for the ST-like toxin EAST1 in EAggEC diarrhea (21), and a 120-kDa protein that cross-reacted with hemolysin antibodies (1) has not been shown to have enterotoxic properties. Here we present data suggesting that a 108-kDa protein secreted by EAggEC strains is a heat-labile enterotoxin. This protein is recognized by sera from patients in an outbreak of EAggEC diarrhea.

The following data suggest that the 108-kDa protein is an enterotoxin: (i) fractions containing both the 108-kDa protein and a distinct 116-kDa protein produce rises in Isc, whereas a fraction from a strain producing only the 116-kDa protein does not; (ii) polyclonal antibodies raised against the 108-kDa protein abolish enterotoxic activity in a dose-related fashion, whereas anti-116-kDa protein antibodies have no effect; (iii) a 108-kDa protein-encoding subclone from the pAA plasmid induces increases in Isc; a pAA-cured EAggEC strain does not.

The 108-kDa toxin appears to induce not only enterotoxic effects but also tissue damage, inflammation, and mucus secretion; these effects correlated with a fall in R value. These data are consistent with other reports that EAggEC strains elaborate one or more cytotoxins and induce damage to the intestinal mucosa (10, 16, 24).

Thus, our data indicate that at least some EAggEC strains secrete a high-molecular-mass (ca. 108-kDa) protein which is encoded on the pAA virulence plasmid and has enterotoxic and perhaps cytotoxic activity on intestinal preparations. The enterotoxic effects were characterized by an increase of Isc and PD and a decrease in R, indicating induction of a net secretory state and damage to epithelial cells and/or their cellular junctions. This enterotoxin is immunogenic, as antibodies against the 108-kDa protein can be found in sera from children with EAggEC infection. The 108-kDa enterotoxin could play an important role in the diarrhea produced by EAggEC.

Our data allow us to hypothesize a model of EAggEC infection in which initial adherence is mediated by AAF fimbriae, followed by the induction of a net secretory state induced by the 108-kDa enterotoxin and also perhaps EAST1. This may be followed by the development of cytotoxicity on the mucosa also induced by the 108-kDa toxin or by an as yet unidentified factor. Studies to identify these factors are ongoing.