Porphobilinogen deaminase has been purified and crystallized from an overproducing recombinant strain of Escherichia coli harbouring a hemC containing plasmid which has permitted the isolation of milligram quantities of the enzyme. Determination of the molecular weight of the enzyme by SDS gel electrophoresis (35,000) and gel filtration (32,500) agrees with the gene derived molecular weight of 33,857. The enzyme has a of 19fJ.M, an isoelectric point of 4.5 and an N terminal sequence NH2- MLDNVLRIAT. The enzyme is shown to contain a novel dipyrromethane cofactor which is linked covalently to the protein. The structure of the cofactor is proposed on the basis of its reaction with Ehrlich's reagent and from its chemical properties. Specific labelling of the dipyrromethane cofactor by growth of the E. coli in the presence of 5-amino[5- 14Cjlaevulinic acid demonstrated that the cofactor is not subject to catalytic turnover. The structure of the cofactor was further confirmed as a dipyrromethane made up of two linked pyrrole rings by [13C] n.m.r. studies after the deaminase was specifically labelled with [13C] by growth of the bacteria on 5-amino[5-13C]laevulinic acid. The chemical shift data indicate that one of the pyrrole rings of the cofactor is covalently attached to the deaminase enzyme through a cysteine residue. Evidence from protein chemistry studies confirm that cysteine-242 is the covalent binding site for the cofactor. The formation of the dipyrromethane cofactor of E. coli was shown to depend on the presence of 5-aminolaevulinic acid. A hemA" mutant formed inactive porphobilinogen deaminase when grown in the absence of 5-aminolaevulinic acid since this strain was unable to biosynthesise the dipyrromethane cofactor. The mutant formed normal levels of deaminase, however, when grown in the presence of 5 aminolaevulinic acid. Porphobilinogen, the substrate, interacts with the free a-position of the dipyrromethane cofactor to give three stable enzyme-intermediate complexes, ES, ES2 and ES3. Experiments with regiospecifically labelled intermediate complexes have shown that, in the absence of further substrate molecules, the complexes are interconvertible by the exchange of the terminal pyrrole ring of the complex. The formation of enzyme-intermediate complexes is accompanied by the exposure of a cysteine residue suggesting that substantial conformational changes occur on binding substrate. Experiments with the a-substituted substrate analogue, abromoporhobilinogen, have provided further evidence that the cofactor is responsible for the covalent binding of the substrate at the catalytic site. Based on these findings it has been possible, for the first time, to construct a mechanistic scheme for the deaminase reaction involving a single active site which is able to catalyse the addition or removal of either NH3 or H2O. The role of the cofactor as both a primer and as a means forregulating the number of substrates bound in each catalytic cycle is discussed.