Characterization of membrane translocation by anthrax protective antigen

Solving the crystallographic structure of the ring-shaped heptamer formed by protective antigen (PA), the B moiety of anthrax toxin, has focused attention on understanding how this oligomer mediates membrane translocation of the toxin's A moieties. We have developed an assay for translocation in which radiolabeled ligands are bound to proteolytically activated PA (PA(63)) at the surface of CHO or L6 cells, and translocation across the plasma membrane is induced by lowering the pH. The cells are then treated with Pronase E to degrade residual surface-bound material, and protected ligands are quantified after fractionation by SDS-PAGE. Translocation was most efficient (35%-50%) with LFN, the N-terminal PA binding domain of the anthrax lethal factor (LF). Intact LF, edema factor (EF), or fusion proteins containing LFN fused to certain heterologous proteins [the diphtheria toxin A chain (DTA) or dihydrofolate reductase (DHFR)] were less efficiently translocated (15%-20%); and LFN fusions to several other proteins were not translocated at all. LFN with different N-terminal residues was found to be degraded according to the N-end rule by the proteasome, and translocation of LFN fused to a mutant form of DHFR with a low affinity for methotrexate (MTX) protected cells from the effects of MTX. Both results are consistent with a cytosolic location of protected proteins. Evidence that a protein must unfold to be translocated was obtained in experiments showing that (i) translocation of LF(N)DTA was blocked by introduction of an artificial disulfide into the DTA moiety, and (ii) translocation of LFNDHFR and LF(N)DTA was blocked by their ligands (MTX and adenine, respectively). These results demonstrate that the acid-induced translocation by anthrax toxin closely resembles that of diphtheria toxin, despite the fact that these two toxins are unrelated and form pores by different mechanisms.