Comparison of nucleotide loss from the miscellaneous category of products in YW112, rad27, and the complemented strain. A histogram is shown for double-strand break junctional sequences, showing the number of ends along the y axis and the number of nucleotides lost on the x axis. N is the total number of ends scored for each histogram. When the precise number of nucleotides lost from an end was ambiguous because of microhomology use at the junction, the nucleotide loss from each end was calculated as half of the nucleotides lost from the junction; then, the loss was evenly distributed among all potential positions within each end.

Models for nonhomologous DNA end joining involving FEN-1 (RAD27). (A) Double-strand DNA breaks can have 5′ or 3′ overhangs or can be blunt. The two ends must be brought together into synapsis. This provides an opportunity for microhomology search of one DNA end by the other, resulting in end alignment. If 5′ flaps are generated by this alignment, then FEN-1 (called RAD27 in S. cerevisiae) is the predominant nuclease to remove the flaps (open arrows), as part of the end processing in preparation for ligation by DNA ligase IV. (B) Same as in A, except as follows. The nucleolytic removal of nucleotides from overhangs occurs by exonucleases in a nonprocessive (one at a time) fashion. A subset of these end-processing events happen to anneal, resulting in alignment, followed by ligation. A nonprocessive removal of one nucleotide at a time from each end would generate a diversity of outcomes, each with a small percentage contribution to the diverse array of possible products. This is not consistent with the 60% flap usage seen in Table 2. Moreover, such a pathway would not be affected by deletion of RAD27 because the FEN-1 nuclease, including RAD27 in S. cerevisiae, is not active on single-stranded 5′ ends or blunt ends (21, 31, 32).

Substrate structure and possible recircularization products. pBX2 is a centromeric plasmid with a URA3 selectable marker and carries the ADE2 gene as a reporter. It is derived from pES16. To destroy the XhoI site in the polylinker, pES16 was digested with SalI–XhoI and recircularized. Two PCR fragments (NotI–XhoI and XhoI–BglII) were ligated between the BglII site of the ADE2 gene and the NotI site in the polylinker, resulting in the junction sequence as shown. The addition of two AA nucleotides shifts the ADE2 gene ORF (shown by two arrows). BamHI–XhoI-linearized pBX2 with the indicated 5′-overhanging ends was transformed into yeast. The two ends are joined intracellularly to produce different types of end joints (lower portion of figure). A flap structure is formed by first annealing CT to GA, followed by flap cleavage. A blunted structure is the result of four-nucleotide fill-in synthesis on each end (fill-in nucleotides shown in bold). A gap structure is produced by first annealing AG to TC, then fill-in synthesis of the gap on each side (bold nucleotides). Miscellaneous products are made by trimming each end to varying extents, followed by apparent blunt ligation. The flap, blunted, and some of the miscellaneous products yield an ORF of the ADE2 gene to yield white colonies (ADE2 yeast). The gap and most miscellaneous products, as well as the starting plasmid, do not generate an ORF of the ADE2 gene; hence, they generate red colonies (ade2 yeast). A small fraction of the plasmid was uncut or was cut by only one of the two restriction enzymes (BamHI or XhoI), and these yield red colonies with the sequence shown at the bottom of the figure.