Based on the 3'-and 5'-terminal sequences of DNA of phage X, P2, and 186, a model is proposed for recognition of DNA sequences by enzymes responsible for generation of cohesive ends. Two copies of the cohered ends, either on separate molecules or on a concatemer, are aligned with their helical axes parallel but running in opposite directions. The nicking system is dimeric, with each of the two monomers carrying identical sequencerecognition sites. Two pairs of nicks are introduced into the two aligned DNA molecules by the nicking system. The applicability of this model to other biological processes, such as integration of a viral genome into a host genome and the cutting of concatemeric T7 DNA, is discussed.For several phages, the DNA extracted from mature phage has short, single-stranded ends with complementary base sequences (1). For coliphages, it appears that there are only two classes of cohesive ends, with those of the lambdoid phages (X, 480, 21, 82, 424, and 434) in one class and those of the 186 family (including 186, P2, P4, and 299) in another (1, 2). Within each class base sequences of the ends are identical or very similar. The base sequences of X-type cohesive ends and 186-type cohesive ends are different, and mutual joining does not occur. Shortly after infection, the cohesive ends join and the two single-chain interruptions in the resulting molecule are subsequently sealed by ligase (1). Therefore, in order to regenerate the single-stranded ends before maturation of the phage, whether from a concatemer or from a circular DNA, two staggered single-chain scissions must be introduced at unique points of the DNA molecule. The enzyme or enzymes involved in this process has been named the termini-generating enzyme or Ter for short (3, 4). For X, it appears that the phage gene-A product provides the Ter function, although participation of host functions and X genes between R and A has not been ruled out (5).We studied the problem of DNA sequence recognition involved in the Ter system, aside from interests in this system itself, because the DNA sequences flanking the sites of endonucleolytic nicking are present at the 3' and 5' termini of mature phage DNA; therefore, the base sequences can be determined without the formidable difficulties of determining the sequence of a segment in the middle of a DNA molecule. Elegant work on determination of the sequence of the 5' single-stranded ends has been done by Wu et al. (6,7) and by Murray and Murray (2). Information on the 3'-termini sequences comes primarily from the work of Weigel et al. (8) and our own effort (ref. 9; D. P. Brezinski and J. C. Wang, unpublished). 2667 In this communication, we postulate a model for recognition of DNA base sequences by Ter, based on sequence information. We believe that sequence recognition in the Ter system involves alignment of two copies of the cohered ends with their helical axes parallel but running in opposite directions. We further postulate that the alignment is achieved by a sequence-specific protein s...