Forssman synthetase expression results in diminished shiga toxin susceptibility: a role for glycolipids in determining host-microbe interactions

Abstract

Forssman glycolipid (FG), the product of Forssman synthetase (FS), is widely expressed among nonprimate mammalian species. Here, we describe a molecular and genetic relationship between FG expression and Shiga toxin (Stx) susceptibility. We have isolated the FS cDNA from human, canine, and murine cells. Whereas the murine and canine FS genes express a functional enzyme, the human FS cDNA was found to express a protein that lacks FS activity, despite a high degree of sequence identity with the enzymatically active murine and canine FS genes. In order to examine the relationship between FG expression and Stx susceptibility, Vero cells were transfected with the three FS orthologues or a vector control. Complementation with the human FS cDNA had no effect on Stx susceptibility, whereas stable expression of the canine and murine FS resulted in markedly decreased susceptibility to toxin. Among individual cells, an inverse correlation between FG expression and Stx binding was demonstrated. Moreover, only strongly FG-reactive cells were capable of growing in the presence of Stx. These cells were found to have high levels of FG expression and a correspondingly diminished GbO(3) content. We conclude that expression of a functionally active FS modifies Stx receptor glycolipids to FG and results in markedly decreased susceptibility to toxin. We speculate that inactivation of the FS gene during primate evolution may account, at least in part, for the marked susceptibility of human cells to Stx.

FIG. 1.

Sequence comparison of FS orthologues. (A) The deduced peptide sequences of canine, murine, and human FS proteins are aligned. Identical residues are shaded and boxed, while similar residues are lightly shaded. (B) Schematic representation of the relatedness of FS orthologues. The canine, murine, and human FS peptide sequences were compared by using ClustalW, and the output was plotted by unrooted bootstrap analysis.

FIG. 2.

Vero cells expressing a functional FS are resistant to Stx. Cells stably transfected with the human, truncated canine (wild-type), murine, or canine FS were incubated with various concentrations of Stx and pulsed with [³⁵S]methionine, and the amount of incorporated radiolabel was determined by TCA precipitation and scintillation counting. Data represent the mean of triplicate samples.

FIG. 3.

Stx binding correlates inversely with FG expression. Wild-type Vero cells or those stably transfected with the canine FS were dual labeled with AlexaFluor 488 StxB (green) or anti-Forssman antibodies and AlexaFluor 568 (red)-conjugated secondary antibody. Labeled cells were visualized by confocal immunofluoroscopy and FACS. For each cell type, two immunofluorescence panels are shown above the FACS profile of the same cells. The FACS data are presented with Stx1 reactivity plotted on the x axis and anti-FG reactivity on the y axis. Binding of both StxB and anti-FG would shift cells to the upper right quadrant, whereas a shift to the upper left or lower right quadrants indicates binding of only anti-FG or StxB, respectively.

FIG. 4.

Exposure to Stx maintains high-level FG expression. (A) Vero cells stably transfected with the canine FS were passaged serially either in the absence (− toxin) or presence (+ toxin) of Stx and then dual labeled with AlexaFluor 488-conjugated StxB and anti-Forssman antibodies as described above. Cells were imaged by using confocal immunofluoroscopy and FACS. (B) Vero cells stably transfected with the canine FS and passaged in toxin were dual labeled with anti-Forssman antibodies (red) as above and AlexaFluor 488-conjugated cholera toxin B-subunit. Boxed numbers in the FACS profiles indicate the percentage of cells in the upper right quadrant, representing the percentage of cells bound by both anti-FG and labeled toxin (StxB or cholera toxin).

FIG. 5.

Cells expressing high levels of FG fail to bind Stx due to decreased GbO₃ content. (A) Wild-type Vero cells or those transfected with the canine FS and passaged in Stx were incubated with various concentrations of ¹²⁵I-labeled StxB (total) or excess unlabeled toxin (nonspecific). After washing, the amount of bound [¹²⁵I]StxB was quantitated. Data represent the mean of duplicate samples. (B) Crude lipid extracts were prepared from wild-type Vero cells (WT) or FS-transfected cells exposed to Stx (FS+). After separation by TLC, GbO₃ or FG was detected by monoclonal antibody followed by peroxidase-conjugated secondary antibody and ECL. (C) Glycolipids were extracted from 2 × 10⁵ untransfected (“C”) Vero cells or those transfected with the human FS (WT) or canine FS exposed to Stx (FS+). After separation by TLC, extracts were overlaid with ¹²⁵I-labeled Stx1. After washing, bound toxin was detected by autoradiography. Migration of glycolipids standards is indicated by the arrows. CDH, ceramide dihexoside.