RecBCD enzyme is a multifunctional nuclease that is essential for the major pathway of homologous genetic recombination in Escherichia coli. It has a potent helicase activity that uses ATP hydrolysis to unwind very long stretches of DNA. The functional form of RecBCD enzyme has been unclear, since M r of 250,000 -655,000 have been previously reported. We have isolated two oligomeric forms of the enzyme, one (monomeric) containing a single copy of the RecB, RecC, and RecD polypeptides, and the other (dimeric) containing two copies of each polypeptide. We show here that the monomeric form of the enzyme (M r Ϸ 330,000) can form a stable initiation complex on the end of ds DNA. Depending on the nature of the ds end, K D estimates ranged from Ϸ 0.1 nM to Ϸ 0.7 nM in the presence of Mg 2؉ ions, which enhanced but was not required for binding. We further showed that the complex of monomeric RecBCD enzyme and a ds DNA end was competent to unwind DNA. A general model for the action of helicases has been proposed that uses repeated conformational changes between two states of a complex between DNA and a dimeric form of the enzyme. Our results make such a model unlikely for RecBCD enzyme.The RecBCD enzyme (EC 3.1.11.5) is a large ATP-dependent enzyme that is involved in recombination and repair of DNA in Escherichia coli (reviewed in Ref. 1). It is encoded by the recB, recC, and recD genes, whose gene products have M r of 134,000, 129,000, and 67,000, respectively, as inferred from DNA sequence data and N-terminal peptide analysis (2-5). The enzyme has a potent ATP-dependent exonuclease that is active on either ds 1 or ss DNA and a weak ATP-stimulated endonuclease activity that acts only on ss DNA. It can use the energy of ATP hydrolysis to unwind ds DNA, either transiently or permanently, in a highly processive reaction (6, 7). The enzyme is active on linear, but not circular, ds DNA and thus requires a ds terminus for its unwinding or nuclease activity (8, 9).Published reports suggest that RecBCD enzyme can exist in either a monomeric (B 1 C 1 D 1 ) or a dimeric (B 2 C 2 D 2 ) form.2 The native M r of the enzyme was initially reported to be about 250,000 (9), as estimated by glycerol gradient centrifugation, consistent with the enzyme molecule containing one copy each of the RecB, RecC, and RecD polypeptides (with a predicted M r of 330,000). A higher M r form of the enzyme, apparently dimeric, was observed in sonicates of E. coli (10), together with the previously reported form, but was lost during subsequent purification steps. RecBCD enzyme purified from a strain that overproduced the enzyme had a native M r of 655,000 but which decreased to about 270,000 in the presence of 0.5 M NH 4 Cl (11). The higher M r form was not observed in a subsequent purification (12). More recently, RecBCD enzyme has been produced either by overproduction of the three subunits within E. coli or by mixing of purified subunits (5). M r estimations by gel filtration or native polyacrylamide gel electrophoresis are consistent with the enzyme ...