1. Biochemistry Department, University of Leicester, Leicester LE1 7RH, UK
2. The Burnham Institute, 10901 North Torrey Pines Road, La Jolla, California 92037, USA
3. Dana Farber Cancer Institute, Boston, Massachusetts 02115, USA
4. Department of Microbiology and Molecular Genetics, Harvard Medical School, 200 Longwood Avenue, Boston, Massachusetts 02115, USA
5. National Institute of Dental and Craniofacial Research, National Institutes of Health, 9000 Rockville Pike, Bethesda, Maryland 20892, USA
6. Department of Microbiology and Immunology, University of Michigan Medical School, 5641 Medical Science II, Ann Arbor, Michigan 48109, USA
7. Present addresses: AstraZeneca, Alderley Park, Macclesfield SK10 4TG, UK (A.P.); EMBL Grenoble Outstation, 38042 Grenoble Cedex 9, France (C.P.); Boston University, 36 Cummington Street, Boston, Massachusetts 02215, USA (J.B.).

Correspondence to: Robert C. Liddington³ Correspondence and requests for materials should be addressed to R.C.L (e-mail: Email: rlidding@burnham.org). Coordinates for the monoclinic LF crystal structure and the cubic LF–MAPKK2 complex have been deposited with the Protein Data Bank (accession codes 1J7N and 1JKY, respectively).

Lethal factor (LF) is a protein (relative molecular mass 90,000) that is critical in the pathogenesis of anthrax^{1, 2, 3}. It is a highly specific protease that cleaves members of the mitogen-activated protein kinase kinase (MAPKK) family near to their amino termini, leading to the inhibition of one or more signalling pathways^{4, 5, 6}. Here we describe the crystal structure of LF and its complex with the N terminus of MAPKK-2. LF comprises four domains: domain I binds the membrane-translocating component of anthrax toxin, the protective antigen (PA); domains II, III and IV together create a long deep groove that holds the 16-residue N-terminal tail of MAPKK-2 before cleavage. Domain II resembles the ADP-ribosylating toxin from Bacillus cereus, but the active site has been mutated and recruited to augment substrate recognition. Domain III is inserted into domain II, and seems to have arisen from a repeated duplication of a structural element of domain II. Domain IV is distantly related to the zinc metalloprotease family, and contains the catalytic centre; it also resembles domain I. The structure thus reveals a protein that has evolved through a process of gene duplication, mutation and fusion, into an enzyme with high and unusual specificity.