A 4-kb HindII frgment that supported the efficient autonomous replication of plasmid vector pDY-, a replication-defective construct based on Epstein-Barr virus sequences, in human K562 cells was rescued from amplified double-minute chromosomes containing the murine adenosine deaminase locus. Polymerase chain reaction assays of size-fractionated nascent strands demonstrated that replication initiation occurred within the same 1-to 2-kb region of this fragment in autonomously replicating plasmids containing the sequence in either orientation, in double-minute chromosomes, and in the single-copy locus at its normal chromosomal location. The complete sequence of this fragment was determined; it contains a 248-bp polypurine tract and consensus binding site sequences for several putative transcription and replication factors.Our understanding of the molecular mechanisms of DNA replication in mammalian cells is largely based on the study of viruses. Although such studies have been extremely productive, there are certain aspects of chromosomal replication that cannot be addressed with such model systems. For example, although the initiation of viral replication is clearly dependent on specific sequences that are recognized by self-encoded proteins, such as simian virus 40 T antigen, neither the DNA signals nor the proteins involved in the initiation of replication in mammalian chromosomes have been defined. In addition, there are unique temporal and spatial controls that must come into play with chromosomal replication to coordinate the accurate and orderly duplication of each DNA base in the entire genome during the S phase.It is clear that for the replication of mammalian chromosomes during the S phase at the measured rates of replication fork movement, the initiation of DNA synthesis must occur at many sites. By analogy with prokaryotic systems (40, 44) and simple eukaryotic systems (10,11,24,42,43,51,80), one would expect this initiation to be dependent on specific signals. However, the continuing ambiguity about the identity of such signals must be due to one of two factors: either (i) they do not exist or (ii) the intrinsic complexity and/or redundancy of these signals make them difficult to dissect by standard approaches. The fact that many of the assays available to search for such sequences are both experimentally difficult and dependent on fundamental assumptions concerning the properties of putative replication intermediates further complicates this issue (74). The most promising candidates for mammalian chromosomal replication origins are sites flanking the DHFR locus in Chinese hamster cells (4, 12-15, 38, 53, 73, 75), and the c-myc locus in the human genome (55, 72), in which sequences have been implicated in origin activity by multiple assays. However, even at the most rigorously studied region 3' of the DHFR gene, the conclusions from different approaches have not been completely consistent; whereas some assays indicate sequence specificity in origin activity (14,37,73) (27,75). Models for the reconcilia...