The local base preferences surrounding 5' ends of HIV-1 proviruses integrated in ATM-deficient cells. (A) A schematic figure of the strand transfer reaction of HIV-1. The 5' end of viral DNA attacks the phosphodiester bond between positions -1 and 1 of the host DNA, and covalently joins to the position 1 nucleotide. (B) Base compositions around the integration sites in the control MRC5SV cells and ATM-deficient cells. The sequences represent the target DNA sequence before the viral DNA is inserted between the position 1 and -1. The 5 bp sequences (positions 1 to 5), which are duplicated next to both ends of the provirus, are boxed by blue lines. Each tabulated number represents the observed base frequency divided by the expected base frequency at each position. The expected base frequencies are average frequencies observed in the human genome (A:29%, G:21%, C:21%). The P values are obtained by χ2 analysis comparing observed and expected base compositions at each position. At the positions with P < 0.01, frequencies < 60% and frequencies > 140% of expected frequencies are colored yellow and green, respectively.