The Cholera toxin project

"Lunar Landers" - A model of the heat-labile enterotoxin (LT) from E. coli attacking a gut cell. The two lunar lander modules are each a single hexameric protein assembly as seen in the crystal structure of the LT AB5 holotoxin. The lunar surface is the outer membrane of an intestinal epithelial cell. Just visible protruding from the membrane surface under the toxin which has landed are five copies of the saccharide moitie of ganglioside GM1, to which the toxin binds. GM1 is a normal membrane component which the toxin coopts as a receptor.

The toxin project was spearheaded by Wim Hol. The drug design component was a collabortive effort by groups at the UW Biomolecular Structure Center ( Ethan Merritt, Christophe Verlinde, Erkang Fan) and elsewhere.

Introduction

Cholera toxin is an AB5 hexameric assembly secreted by Vibrio cholerae. As with many other bacterial toxins the catalytic activity resides in the `A' fragment, in this case a separate subunit, while receptor binding and delivery of the toxin to the target cell is mediated by a separate `B' fragment, in this case a pentamer. The class of AB5 toxins may be subdivided into families based on sequence homology and catalytic activity. The cholera toxin family includes in addition to cholera toxin itself the E. coli heat-labile enterotoxins LT and LT-II. The closely-related shiga toxin family comprises a number of toxins from Shigella dysenteriae and the `shiga-like' toxins (also known as verotoxins) from E. coli. The effect of these toxins on human populations ranges from the relatively mild travelers' diarrhea caused by infection with E. coli strains producing LT to the acute and life-threatening diarrhea caused by V. cholerae infection and the equally serious hemolytic uremic syndrome (`hamburger disease') caused by members of the shiga toxin family. Together these AB5 toxins are responsible for over a million deaths annually.

Why we are studying these toxins?

Questions of basic science
How does the toxin recognize and bind to the cell surface? How does it then enter the cell? What about the catalytic mechanism?
Drug design
We use X-Ray crystallography as a tool for structure-based drug design targeting multiple sites on the toxins. Current approaches include blocking the receptor-binding sites on the B-pentamer, blocking the catalytic site on the A subunit, and interfering with holotoxin assembly.
Vaccine design
Because these toxins stimulate the mucosal immune system, there is great interest in using an engineered form of the toxin as a basis for the design of vaccines against a wide range of diseases.

Some Results

Active site mutants
E. coli heat-labile enterotoxin We have structures for a number mutant toxins containing various single site mutantations near the active site of the heat-labile enterotoxin. The Arg7Lys substitution [1995f] leads to substantial destabilization of the loop comprised of residues 47-56, a possible indication of conformational changes which occur during substrate binding. The Val97Lys substitution [1995b] destroys catalytic activity but does not induce a significant conformational change at the active site. The larger Lys sidechain occupies an interior cavity, displacing several water molecules seen in the wildtype toxin. The Lysine sidechain also forms a salt bridge to the catalytic residue Glu 112. The structure of a partially activated form of the heat-labile enterotoxin, in which the A1 and A2 domains of the catalytic subunit are separated by proteolytic cleavage but remain linked by a disulfide bond, showed very little change from the fully intact holotoxin [1994c]. This strongly hints that reduction of the disulfide bond is a crucial step in activation of the toxin.
Receptor binding - crystallography and drug design
GM1 binding site The receptor for cholera toxin and heat-labile enterotoxin is ganglioside GM1. The toxin recognizes the branched pentasaccharide portion of GM1 which protrudes from the exterior membrane surface. The structure of the cholera toxin B-pentamer complexed with the complete GM1 pentasaccharide [1994a, 1997d, 1998d] gives us a view of the toxin:receptor binding mode at atomic resolution. It provides both a better understanding of the attachment of the toxin to the cell (see figure at top of page) and a starting point for the structure-based design of receptor binding inhibitors. We have futher explored the stereochemistry of the receptor binding site via complexes of the heat-labile enterotoxin with smaller fragments of the receptor. These studies provide a starting point for iterative structure-based design of potential drugs which block receptor binding [1995e, 1997d]. pentapus schematic We have used structure-guided synthesis [1999c] and combinatorial chemistry [1999d] starting from galactose to find derived compounds that bind to the toxin with an affinity gain of several orders of magnitude relative to galactose itself. A further dramatic increase in receptor-blocking power comes from linking five of these small compounds together to form a "pentapus" whose 5-fold symmetry matches that of the toxin B-pentamer [2000a]. Additional combinatorial, synthetic and structural work followed.
Protein engineering and vaccine design
In order to use the toxin as a vaccine component it is obviously desirable to reduce or abolish the cytotoxicity while retaining the toxins ability to stimulate the immune system. To this end we have undertaken protein engineering of the catalytic A subunit. Based on analysis of the residues likely to play a role in substrate binding, we selected an initial set of four residues near the active site for amino-acid substitution. Initial success has come from cloning and expressing several of these mutants which prove to have reduced cytotoxicity while retaining immunogenicity [Feil et al 1996a, de Haan et al 1998a,b,c].

This work was supported by the NIH (AI34501, AI44954, GM54618) and the University of Washington Royalty Research Fund