Cyclic AMP receptor protein-dependent repression of heat-labile enterotoxin

Abstract

Enterotoxigenic Escherichia coli is a major cause of acute diarrheal illness worldwide and is responsible for high infant and child mortality rates in developing nations. Two types of enterotoxins, one heat labile and the other heat stable, are known to cause diarrhea. The expression of soluble heat-labile toxin is subject to catabolite (glucose) activation, and three binding sites for cAMP receptor protein (CRP or CAP) were identified upstream and within the toxin promoter by DNase I footprinting. One CRP operator is centered at -31.5, thus encompassing the promoter's -35 hexamer. Potassium permanganate footprinting revealed that the occupancy of this operator prevents RNA polymerase from forming an open complex in vitro. However, the operator centered at -31.5 is not sufficient for full repression in vivo because the deletion of the other two CRP binding sites partially relieved the CRP-dependent repression of the heat-labile toxin promoter. In contrast to heat-labile toxin, CRP positively regulates the expression of heat-stable toxin. Thus, the conditions for the optimal expression of one enterotoxin limit the expression of the other. Since glucose inhibits the activity of CRP by suppressing the pathogen's synthesis of cyclic AMP (cAMP), the concentration of glucose in the lumen of the small intestine may determine which enterotoxin is maximally expressed. In addition, our results suggest that the host may also modulate enterotoxin expression because cells intoxicated with heat-labile toxin overproduce and release cAMP.

FIG. 1.

Heat-labile toxin expression is repressed by CRP. All strains were cultured aerobically at 37°C. The values shown within each panel are the means and standard deviations of 6 or 10 independent cultures, collected over 2 or more days. A Student t test was used to calculate P values. (A) ETEC strain H10407 was cultured in phosphate-buffered LB medium (pH 7.5) with or without 0.4% (wt/vol) glucose. Culture aliquots were collected at the indicated points along the growth curves, and as shown in panel B, the concentrations of soluble LT-I were determined by GM1-ELISA. I, II, and III correspond to the aliquots shown on the growth curves, with + and − indicating the presence or absence of glucose in the culture medium. (C) K-12 strains with Lac reporters integrated at attB_HK022 were constructed with the promoters of heat-labile toxin LT-I, eltAp, or heat-stable toxin STa, estAp, from H10407. Reporter strains GPM1160 (eltAp::lacZYA), GPM1162 (Δcrp eltAp::lacZYA), GPM1251 (cyaA::kan eltAp::lacZYA), GPM1159 (estAp::lacZYA), and GPM1161 (Δcrp estAp::lacZYA) were cultured in LB medium, and the expression of β-galactosidase was determined by the Miller method. (D) Expression of β-galactosidase in reporter strains GPM1160 and GPM1162 transformed with the CRP expression plasmid pSE186 or vector pHG165. Both strains were cultured in LB medium with 100 μg/ml ampicillin.

FIG. 2.

Locations of CRP binding sites and the transcription start site of eltAp. Numbering is relative to the transcription start site of eltAp, which is represented by a wavy arrow. (A) Representative DNase I footprints of CRP homodimers bound to the coding and noncoding strands of eltAp in the presence of 200 μM cAMP. Two independent DNase I footprinting experiments were done for each strand; therefore, each CRP binding site was visualized four times. For primer extensions, 103 μg of total RNA from strain GPM1160 (eltAp::lacZYA) was used but only 62 μg from strain GPM1162 (Δcrp eltAp::lacZYA). Primer extensions were repeated twice. Lanes labeled GA and TC contain Maxam-Gilbert sequencing ladders. (B) Nucleotide sequence of eltAp from ETEC strain H10407. The −35 and −10 hexamers are bound by rectangles and have been previously characterized by site-directed mutagenesis (50). Each CRP binding site is compared to the CRP consensus sequence, and the nucleotides shown in bold are part of spaced inverted repeats. Over- and underlines indicate the approximate extent of each DNase I footprint.

FIG. 3.

CRP bound at eltA1o prevents the formation of an open complex at the LT-I promoter. The in vitro potassium permanganate footprint of a DNA fragment carrying eltAp, from −91 to +154 (eltA1o⁺ ΔeltA2o ΔeltA3o), with the noncoding strand radiolabeled is shown. Hyperreactive nucleotides within the RNA polymerase open complex are indicated by black dots. The final concentrations of CRP, cAMP, and RNA polymerase were 100 nM, 500 μM, and 50 nM, respectively. Numbering is relative to the eltAp transcription start site, which is designated by a wavy arrow. The ability of CRP to inhibit open complex formation in the presence of cAMP was visualized in three independent potassium permanganate footprinting experiments of which one representative gel image is shown. Lanes labeled GA and TC contain Maxam-Gilbert sequencing ladders. +, presence; Ø, absence.

FIG. 4.

Differential regulation of ETEC enterotoxins by CRP. (A) In the presence of its ligand cAMP, CRP activates the expression of heat-stable toxin STa and represses the expression of heat-labile toxin LT-I. High concentrations of glucose suppress the synthesis of cAMP, rendering CRP inactive. (B) As a result, the expression of the enterotoxins may change relative to one another as the pathogen moves through the small intestine because glucose concentrations are typically higher in the duodenum than in the ileum. CRP without its cAMP cofactor is represented as CRP_apo.